AUTOMATION HISTORY

Grab is preparing to launch artificial intelligence-powered robots “very soon” to help delivery drivers pick up meal orders from restaurants more quickly, according to Grab Co-founder and Group CEO Anthony Tan. Complete the form to unlock this article and enjoy unlimited free access to all PYMNTS content — no additional logins required. yesSubscribe to our daily newsletter, PYMNTS Today. By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. Δ The robot, dubbed “Carri,” is one of several AI-powered solutions unveiled by Tan in a speech at GrabX 2026, the company’s annual product event. Grab is a super app that operates in Southeast Asia and offers food and grocery deliveries, ride-hailing services and digital financial services. Carri is designed to assist drivers, not replace them, Tan said during the speech. “Our drivers lose around 10% of their earning time looking for a restaurant in a mall or waiting for their customers to come down from office towers,” Tan said. “If this little fellow can help handle that ‘wait’ by finding the restaurant and passing the order to the driver, it allows our drivers to move to the next job more quickly.” PYMNTS reported in March that physical AI is drawing investor attention as venture funding flows toward companies building systems designed to operate in the physical world. On Wednesday (April 15) it was reported that Barclays said robots and drones could reduce food delivery costs to as little as $1 per order. Advertisement: Scroll to Continue Together with Carri, Grab unveiled 13 AI-powered experiences at GrabX 2026, the company said in an April 8 press release. For consumers, Grab introduced Group Ride, which helps riders save on fares by sharing a ride with others; GrabMore, which allows customers to have orders from two merchants delivered with a single delivery fee; Grab Shopping Agent, which is part of Grab AI Assistant; GrabMaps for Consumers, which integrates a Journey Planner and other tools with maps; and Cash Loan, which offers credit to consumers who have limited financial history. For travelers, the company unveiled Personalised Travel Experiences, an AI travel companion that shares information and reminders; GrabStays, which serves as a hotel booking service; Discover by Grab, which provides a guide to local restaurants; and GrabPay for Travel, which aids with payments across Southeast Asia. For merchants and drivers, Grab announced Virtual Store Manager, which uses CCTV hardware to provide information to managers; Cloud Printer, which automatically prints orders for kitchen staff; Tap to Pay, which turns GrabMerchant-enabled smartphones into contactless payment terminals; and Driver AI Assistant, which answers drivers’ questions. “We believe that everyone — regardless of their technical skill — should have the opportunity to jump on this AI wave, not be swept away by it,” Tan said during his speech. “And we want to help as many people as we can.” For all PYMNTS AI coverage, subscribe to the daily AI Newsletter. Delivery Robots Lead Grab’s AI Expansion Circle Chief Says China Could Issue Stablecoin in 3 to 5 Years Amex Acquires Hyper to Boost AI and Expense Management Offerings Anthropic Ready to Offer Mythos to British Banks Get PYMNTS Today, AI, B2B and more. We’re always on the lookout for opportunities to partner with innovators and disruptors.

Ryan Gariepy has been building Canadian robots for nearly two decades. He has also been an outspoken advocate for how they should and should not be used in a military context. Gariepy co-founded Kitchener-Waterloo’s Clearpath Robotics in 2009, co-leading the company as CTO until its acquisition by US industrial automation giant Rockwell Automation in 2023 at a reported price tag of about $600-million USD. These days, Gariepy works as Rockwell’s vice-president of robotics and chairs the Canadian Robotics Council, with a keen eye towards building up the country’s robot-making industry. BetaKit reporter Josh Scott sat down with Gariepy to unpack his thoughts on the recent killer robot discussions that have been brought to the fore by AI, how robotics can help Canada realize its new defence ambitions, and whether the country is doing enough to capture the opportunity he sees. The following interview has been edited for length and clarity. There are a lot of areas where we could be using robots a lot more in Canada, and with that, there could be a lot more robotics companies in Canada. Canada is very well-positioned to be a global leader in robotics. Something that Canada has going for us is this mix of a physical, industry-powered economy and a very educated and cosmopolitan populace. The vast majority of modern mines are going to be using robots to some degree. The same is true for modern manufacturing of any sort, whether it’s cars, whether it’s food, or whether it’s pharma. But they can always be used more. If you go to any modern plant in Canada, it’s heavily using robotics. But then as you go into the broader supply chain you’re less likely to run into robots there. Robots tend to be concentrated more in the large businesses, not because they can’t help in the small business but because they have more time, possibly more capital, and more ability to take some of that risk. We could ask the Ukrainians. We could also ask anyone who’s had to do a long posting in the Arctic. We have a lot more space, we have a lot fewer people, and our environment is a lot more hostile. That is the perfect place for robotics. As much as we are committed to increasing the size of our military, we’re a small country that does play on the global stage, which means that we will need force multipliers. It’s good to see how robotics has been identified as a sovereign capability. We may not be able to act as quickly as Ukraine did—which basically retooled their entire economy around building drones—but we can use our relationships with Ukraine, we can use our established manufacturing capabilities and our natural resources to modernize very quickly and modernize for the next conflict, as opposed to the past one. I support using robots in the military. Logistics, search and rescue, reconnaissance, or training, all of these areas where robots should probably be used. Even weaponized robots, to some degree, for military purposes, are things that I support. At the same time, it’s very important to have reasonable controls and reasonable certifications around how these systems are used. Ten years ago, there were a lot of conversations where we were saying, “AI is going to make mistakes, and it’s going to be confusing and different, and you’re not going to be able to predict it.” And everyone was like “no, no, that’s not the case.” And now we’re here. Anyone who’s got any sort of media awareness knows that AI makes mistakes, and if your AI is, say, misciting an article, and it’s going to make that mistake, are you sure you want that tool deciding on whether or not to use lethal force? We have a risk problem there. As much as we are committed to increasing the size of our military, we’re a small country that does play on the global stage, which means that we will need force multipliers. There’s also a morality and accountability problem. It’s very important that accountability still lies with a human at some point, and that in the end, you don’t leave people with an out to say, “Oh, it wasn’t me. It was the system that committed that war crime.” The military is the most experienced when it comes to the appropriate and proportional use of force. We really want to make sure that responsibility [and] accountability remains with the military as opposed to allowing people to push that off on some engineer who wrote some code 10 years ago. Over the years, I’ve been part of or peripheral to these discussions. People will use that as a political football. It’s most important to maintain a chain of accountability, certification, understanding, and testing of the technology itself. It’s difficult to understand what has been agreed to and not agreed to because you’ve also got OpenAI adding some noise to the conversation. But you don’t designate a company as a supply chain risk and then also say you’re effectively going to nationalize them. There are political factors at play. RELATED: Rockwell Automation completes acquisition of Clearpath Robotics and its OTTO Motors division On a personal note, I support saying that you should not use the specific kind of technology that Anthropic uses as a key component of autonomous weapons. I would certainly agree that using an LLM for targeting decisions is not the best way forward. I might also suspect that there are things that the Anthropic team knows that we don’t, which cause them to draw this line. I don’t think anyone in their right mind would decide, particularly these days, to pick a fight with the US Department of Defense if they didn’t need to. The thing I’m most excited about is society realizing that robots can help right now. We do not need to wait, and shouldn’t wait until there’s a humanoid knocking on your door to do your laundry. Robots can help, and they can help right now. They can help us be safer, more productive, and more comfortable. The thing that keeps me up is how much opportunity Canada is leaving on the table here. We have an opportunity to build a more secure country, and we’re not moving fast enough. Feature image courtesy Clearpath Robotics. The publication of record for Canadian tech and startup news since 2012. Learn more

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Advertisement Scientific Reports , Article number: (2026) Cite this article We are providing an unedited version of this manuscript to give early access to its findings. Before final publication, the manuscript will undergo further editing. Please note there may be errors present which affect the content, and all legal disclaimers apply. Social and emotional intelligence are fundamental to human cognition, yet current artificial agent frameworks typically treat these capabilities separately, limiting their ability to generate authentic social interactions. We present SELAgents (Social and Emotional Learning Agents), a novel framework that integrates emotional processing, theory of mind, and social learning within a unified reinforcement learning architecture. The framework combines a three-dimensional emotional state space (Pleasure-Arousal-Dominance model), Bayesian belief networks for mental state inference, and game-theoretic social strategy selection. Through systematic experiments with populations of 10 heterogeneous agents over 200 timesteps (30 independent runs), we demonstrate significant improvements over traditional reinforcement learning baselines: emotional intelligence scores increased by 49% (0.73 ± 0.08 vs 0.49 ± 0.11, \(p < 0.001\)), social coherence improved by 66% (0.68 vs 0.41, \(p < 0.001\)), and resource allocation efficiency reached 87% (vs 62% baseline, \(p < 0.001\)). Agents exhibited emergent behaviors including emotional contagion effects (correlation strength \(\gamma = 0.72\) in dense networks) and stable coalition formation (4.3 ± 1.2 agents per coalition). Ablation studies revealed that theory of mind capabilities contributed most significantly to performance (31.2% degradation when removed), followed by emotional processing (28.7%) and social strategies (22.4%). These results suggest that integrating emotional processing with social learning mechanisms produces more sophisticated agent behaviors that exhibit patterns consistent with human social dynamics. We provide our complete implementation as open-source software to facilitate further research. This study assumes perfect observability of emotional states, representing an upper bound on achievable performance; extending the framework to partial observability settings remains an important direction for future work. All data and code used in this study are openly available at: https://github.com/nicolastorresr/SELAgents. We have made our implementation openly available via GitHub to facilitate reproducibility and encourage further research in this domain. Picard, R. W. Affective computing: Challenges. Int. J. Hum Comput Stud. 59(1–2), 55–64 (2000). Google Scholar Adolphs, R. The social brain: Neural basis of social knowledge. Annu. Rev. Psychol. 60, 693–716 (2009). Google Scholar Silver, D. et al. Mastering the game of go with deep neural networks and tree search. Nature 529(7587), 484–489 (2016). Google Scholar Brown, T. et al. Language models are few-shot learners. Adv. Neural. Inf. Process. Syst. 33, 1877–1901 (2020). Google Scholar Dautenhahn, K. Socially intelligent robots: Dimensions of human-robot interaction. Philos. Trans. R. Soc. B: Biol. Sci. 362(1480), 679–704 (2007). Google Scholar Damasio, A. R. Descartes’ Error: Emotion, Reason, and the Human Brain (Putnam Publishing, 1994). Shamay-Tsoory, S. G., Aharon-Peretz, J. & Perry, D. The neural correlates of understanding the other’s distress: A positron emission tomography investigation of accurate empathy. Neuroimage 54(3), 2462–2474 (2010). Google Scholar Loewenstein, G. F., Weber, E. U., Hsee, C. K. & Welch, N. Risk as feelings. Psychol. Bull. 127(2), 267–286 (2001). Google Scholar Baker, C. L., Jara-Ettinger, J., Saxe, R. & Tenenbaum, J. B. Rational quantitative attribution of beliefs, desires and percepts in human mentalizing. Nat. Hum. Behav. 1(4), 1–10 (2017). Google Scholar Hernández-Leal, P., Kartal, B. & Taylor, M. E. A survey and critique of multiagent deep reinforcement learning. Auton. Agent. Multi-Agent Syst. 33(6), 750–797 (2019). Google Scholar Adolphs, R. Cognitive neuroscience of human social behaviour. Nat. Rev. Neurosci. 4(3), 165–178 (2003). Google Scholar Breazeal, C. Emotion and sociable humanoid robots. Int. J. Hum Comput Stud. 59(1–2), 119–155 (2003). Google Scholar Premack, D. & Woodruff, G. Does the chimpanzee have a theory of mind?. Behav. Brain Sci. 1(4), 515–526 (1978). Google Scholar Axelrod, R. The Evolution of Cooperation (Basic Books, 1984). Vicci, H. Emotional intelligence in artificial intelligence: A review and evaluation study. SSRN Electron. J. https://doi.org/10.2139/ssrn.4818285 (2024). Google Scholar Raza, M. A., Farooq, M. S., Khelifi, A. & Alvi, A. Emotion-oriented behavior model using deep learning. arXiv preprint. arXiv:2311.14674 (2023). Marsella, S., Gratch, J. & Petta, P. Computational models of emotion. A Blueprint for Affective Computing-A sourcebook and manual 11(1), 21–46 (2010). Google Scholar Tanevska, A., Rea, F., Sandini, G., Cañamero, L. & Sciutti, A. A socially adaptable framework for human-robot interaction. Front. Robot. AI 7, 121 (2020). Google Scholar Leite, I., Castellano, G., Pereira, A., Martinho, C. & Paiva, A. Empathic robots for long-term interaction: Evaluating social presence, engagement and perceived support in children. Int. J. Soc. Robot. 6, 329–341 (2014). Google Scholar Puccetti, N. A., Villano, W. J., Fadok, J. P. & Heller, A. S. Temporal dynamics of affect in the brain: Evidence from human imaging and animal models. Neurosci. Biobehav. Rev. 133, 104491. https://doi.org/10.1016/j.neubiorev.2021.12.014 (2022). Google Scholar Franchi, E. & Poggi, A. Multi-Agent Systems and Social Networks https://doi.org/10.4018/978-1-61350-168-9.ch005 (2011). Fan, R., Xu, K. & Zhao, J. An agent-based model for emotion contagion and competition in online social media. Physica A 495, 245–259 (2018). Google Scholar Fowler, J. H. & Christakis, N. A. Dynamic spread of happiness in a large social network: Longitudinal analysis over 20 years in the Framingham heart study. Bmj 337 (2008) Lazer, D. et al. Computational social science. Science 323(5915), 721–723 (2009). Google Scholar Leibo, J. Z., Zambaldi, V., Lanctot, M., Marecki, J. & Graepel, T. Multi-agent reinforcement learning in sequential social dilemmas. Proceedings of the 16th Conference on Autonomous Agents and MultiAgent Systems 464–473 (2017) Santos, F. C., Pacheco, J. M. & Lenaerts, T. Evolutionary dynamics of social dilemmas in structured heterogeneous populations. Proc. Natl. Acad. Sci. USA 103(9), 3490–3494 (2006). Google Scholar Jara-Ettinger, J. Theory of mind as inverse reinforcement learning. Curr. Opin. Behav. Sci. 29, 105–110. https://doi.org/10.1016/j.cobeha.2019.04.010 (2019). Google Scholar Baker, C. L., Saxe, R. & Tenenbaum, J. B. Bayesian theory of mind: Modeling joint belief-desire attribution. Proceedings of the Annual Meeting of the Cognitive Science Society 33(33) (2011) Oguntola, I., Campbell, J., Stepputtis, S. & Sycara, K. Theory of mind as intrinsic motivation for multi-agent reinforcement learning. arXiv preprint arXiv:2307.01158 (2023) Rabinowitz, N. et al. Machine theory of mind. In International Conference on Machine Learning 4218–4227 (PMLR, 2018). Myerson, R. B. Graphs and cooperation in games. Math. Oper. Res. 2(3), 225–229 (1977). Google Scholar Camerer, C. F. Behavioral Game Theory: Experiments in Strategic Interaction (Princeton University Press, 2011). Hassanpour, S., Rassafi, A. A., González, V. A. & Liu, J. A hierarchical agent-based approach to simulate a dynamic decision-making process of evacuees using reinforcement learning. J. Choice Model. 39, 100288. https://doi.org/10.1016/j.jocm.2021.100288 (2021). Google Scholar Rhodes, S. L., Crabtree, S. A. & Freeman, J. An agent-based model of hierarchical information-sharing organizations in asynchronous environments. J. Artif. Soc. Soc. Simul. 27(2), 2 (2024). Google Scholar Yang, Y. et al. Multi-agent determinantal q-learning. In International Conference on Machine Learning 10757–10766 (PMLR, 2020). Baker, C. L., Jara-Ettinger, J., Saxe, R. & Tenenbaum, J. B. Emergent reciprocity and team formation from randomized uncertain social preferences. Nat. Commun. 10(1), 1–11 (2019). Google Scholar Salovey, P. & Mayer, J. D. Emotional intelligence. Imagin. Cogn. Pers. 9(3), 185–211 (1990). Google Scholar Phelps, E. A. Human emotion and memory: Interactions of the amygdala and hippocampal complex. Curr. Opin. Neurobiol. 14(2), 198–202 (2004). Google Scholar Russell, J. A. A circumplex model of affect. J. Pers. Soc. Psychol. 39(6), 1161 (1980). Google Scholar Kahneman, D. Thinking, Fast and Slow (Farrar, Straus and Giroux, 2011). Wooldridge, M. An Introduction to MultiAgent Systems 2nd edn. (Wiley, 2009). Gamma, E., Helm, R., Johnson, R. & Vlissides, J. Design Patterns: Elements of Reusable Object-oriented Software (Addison-Wesley, 1995). Mehrabian, A. Pleasure-arousal-dominance: A general framework for describing and measuring individual differences in temperament. Curr. Psychol. 14(4), 261–292 (1996). Google Scholar Kingma, D. P. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014). Glorot, X. & Bengio, Y. Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, JMLR Workshop and Conference Proceedings 249–256 (2010). Hagberg, A., Swart, P. & Schult, D. A. Exploring Network Structure, Dynamics, and Function Using Network (Los Alamos National Lab.(LANL), 2008) Shapley, L. S. A value for n-person games. Contrib. Theory Games 2(28), 307–317 (1953). Google Scholar Broekens, J., DeGroot, D. & Kosters, W. A. Formal models of appraisal: Theory, specification, and computational model. Cogn. Syst. Res. 9(3), 173–197 (2008). Google Scholar Osgood, C. E., Suci, G. J. & Tannenbaum, P. H. The Measurement of Meaning Vol. 47 (University of Illinois Press, 1957). Wasserman, S. & Faust, K. Social Network Analysis: Methods and Applications (Cambridge University Press, 1994). Kaelbling, L. P., Littman, M. L. & Moore, A. W. Reinforcement learning: A survey. J. Artif. Intell. Res. 4, 237–285 (1996). Google Scholar Henderson, P., Islam, R., Bachman, P., Pineau, J., Precup, D. & Meger, D. Deep reinforcement learning that matters. In Proceedings of the AAAI Conference on Artificial Intelligence 32 (2018) Mnih, V. et al. Human-level control through deep reinforcement learning. Nature 518(7540), 529–533 (2015). Google Scholar Lowe, R. et al. Multi-agent actor-critic for mixed cooperative-competitive environments. Advances in neural information processing systems 30 (2017) Rashid, T. et al. QMIX: Monotonic value function factorisation for deep multi-agent reinforcement learning. In Proceedings of the 35th International Conference on Machine Learning. Proceedings of Machine Learning Research (eds. Dy, J., Krause, A.) 80, 4295–4304 (PMLR, 2018). https://proceedings.mlr.press/v80/rashid18a.html Ruxton, G. D. The unequal variance t-test is an underused alternative to student’s t-test and the Mann–Whitney U test. Behav. Ecol. 17(4), 688–690 (2006). Google Scholar Premack, D. Why humans are unique: Three theories. Perspect. Psychol. Sci. 5(1), 22–32 (2010). Google Scholar Newman, M. E. The structure and function of complex networks. SIAM Rev. 45(2), 167–256 (2003). Google Scholar Granovetter, M. S. The strength of weak ties. Am. J. Sociol. 78(6), 1360–1380 (1973). Google Scholar Petrović, V. M. Artificial intelligence and virtual worlds-toward human-level ai agents. IEEE Access 6, 39976–39988 (2018). Google Scholar Brooks, R. A. Elephants don’t play chess. Robot. Auton. Syst. 6(1–2), 3–15 (1990). Google Scholar Pfeifer, R. & Bongard, J. How the Body Shapes the Way We Think: A New View of Intelligence (MIT Press, 2006). Hatfield, E., Cacioppo, J. T. & Rapson, R. L. Emotional contagion. Curr. Dir. Psychol. Sci. 2(3), 96–100 (1993). Google Scholar Download references The authors gratefully acknowledge the research support provided by ANID FONDEF IDeA ID25I10018. Department of Electronics, Universidad Técnica Federico Santa María, Av. Vicuña Mackenna 3939, Santiago, RM, 8940897, Chile Nicolás Torres Search author on:PubMed Google Scholar Conceptualization: N.T.; Methodology: N.T.; Formal analysis and investigation: N.T.; Writing—original draft preparation: N.T.; Resources: N.T. Correspondence to Nicolás Torres. The authors declare no competing interests. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/. Reprints and permissions Torres, N. Social and emotional learning in artificial agents. Sci Rep (2026). https://doi.org/10.1038/s41598-026-48309-5 Download citation Received: 14 November 2025 Accepted: 07 April 2026 Published: 19 April 2026 DOI: https://doi.org/10.1038/s41598-026-48309-5 Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative Advertisement Scientific Reports (Sci Rep) ISSN 2045-2322 (online) © 2026 Springer Nature Limited Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

Loading... You are seeing this because the administrator of this website has set up Anubis to protect the server against the scourge of AI companies aggressively scraping websites. This can and does cause downtime for the websites, which makes their resources inaccessible for everyone. Anubis is a compromise. Anubis uses a Proof-of-Work scheme in the vein of Hashcash, a proposed proof-of-work scheme for reducing email spam. The idea is that at individual scales the additional load is ignorable, but at mass scraper levels it adds up and makes scraping much more expensive. Ultimately, this is a placeholder solution so that more time can be spent on fingerprinting and identifying headless browsers (EG: via how they do font rendering) so that the challenge proof of work page doesn't need to be presented to users that are much more likely to be legitimate. Please note that Anubis requires the use of modern JavaScript features that plugins like JShelter will disable. Please disable JShelter or other such plugins for this domain. Sadly, you must enable JavaScript to get past this challenge. This is required because AI companies have changed the social contract around how website hosting works. A no-JS solution is a work-in-progress. Protected by Anubis From Techaro. Mascot design by CELPHASE. This website is running Anubis version devel.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Advertisement npj Digital Medicine , Article number: (2026) Cite this article 2090 Accesses 9 Altmetric Metrics details We are providing an unedited version of this manuscript to give early access to its findings. Before final publication, the manuscript will undergo further editing. Please note there may be errors present which affect the content, and all legal disclaimers apply. Artificial intelligence (AI) is increasingly embedded in clinical environments, raising questions of trust, fairness, empathy, and governance. The ethical terrain surrounding AI in medicine remains unstable despite its rapid adoption. We introduce the “Seven Deadly Sins of AI in Medicine”, a conceptual framework of recurring systemic failure modes: (i) Blind Trust, (ii) Overregulation, (iii) Dehumanization, (iv) Misaligned Optimization, (v) Overinforming and False Forecasting, (vi) Misapplied Statistics, and (vii) Self-Referential Evaluation. The framework was developed through systematic synthesis of scientific literature, clinical guidelines, and regulatory frameworks prior to any empirical data collection. To validate this pre-established framework, we conducted a global, cross-professional opinion poll of 914 stakeholders from 143 countries between July 2024 and March 2025. Results confirmed broad agreement with each pre-identified risk, revealing cross-cultural convergence in ethical concern alongside persistent divides in attitudes toward regulation—particularly between technologically advanced nations and emerging economies. We further propose an inversion of the framework into seven cardinal virtues for AI in medicine, offering actionable principles to guide responsible development and governance. The goal is to move beyond scattered ethical guidelines toward a unified diagnostic tool for trustworthy, human-centered medical AI. All anonymized data, all figures, and analysis scripts are openly available at (https:/github.com/human-centered-ai-lab/7-sins-of-medical-ai). Code for data visualisation and statistical analysis is available in the same repository under an open-source licence. Rajpurkar, P., Chen, E., Banerjee, O. & Topol, E. J. AI in health and medicine. Nat. Med. 28, 31–38 (2022). Google Scholar Rajkomar, A. et al. Scalable and accurate deep learning with electronic health records. npj Digital Med. 1, 18 (2018). Google Scholar Shick, A. A. et al. Transparency of artificial intelligence/machine learning-enabled medical devices. npj Digital Med. 7, 21 (2024). Google Scholar Paulus, J. K. & Kent, D. M. Predictably unequal: understanding and addressing concerns that algorithmic clinical prediction may increase health disparities. npj Digital Med. 3, 99 (2020). Google Scholar Comeau, D. S., Bitterman, D. S. & Celi, L. A. Preventing unrestricted and unmonitored AI experimentation in healthcare through transparency and accountability. npj Digital Med. 8, 42 (2025). Google Scholar Holzinger, A., Zatloukal, K. & Müller, H. Is human oversight to AI systems still possible?. N. Biotechnol. 85, 59–62 (2025). Google Scholar Klingbeil, A., Grützner, C. & Schreck, P. Trust and reliance on AI: An experimental study on the extent and costs of overreliance on AI. Computers Hum. Behav. 160, 108352 (2024). Google Scholar Saenz, A. D., Harned, Z., Banerjee, O., Abràmoff, M. D. & Rajpurkar, P. Autonomous AI systems in the face of liability, regulations and costs. npj Digital Med. 6, 185 (2023). Google Scholar Akingbola, A., Adeleke, O., Idris, A., Adewole, O. & Adegbesan, A. Artificial intelligence and the dehumanization of patient care. J. Med. Surg. Public Health 3, 100138 (2024). Google Scholar Sharma, D. et al. Triage-Bot: An assistive triage framework. In 2024 IEEE International Conference on Digital Health (ICDH), 138–140 (IEEE, 2024). Caruso, I. et al. Artificial intelligence and the doctor–patient relationship expanding the paradigm of shared decision making. Bioethics 37, 424–432 (2023). Google Scholar Sauerbrei, A., Kerasidou, A., Lucivero, F. & Hallowell, N. The impact of artificial intelligence on the person-centred, doctor–patient relationship: some problems and solutions. BMC Med. Inform. Decis. Mak. 23, 73 (2023). Google Scholar Vamplew, P., Dazeley, R., Foale, C., Firmin, S. & Mummery, J. Human-aligned artificial intelligence is a multiobjective problem. Ethics Inf. Technol. 20, 27–40 (2018). Google Scholar Yoon, Y., Guimaraes, T. & O’Neal, Q. Exploring the factors associated with expert systems success. MIS Q. 19, 83–106 (1995). Google Scholar Mahajan, A. & Gilbert, S. Do we need AI guardians to protect us from health information overload?. npj Digital Med. 8, 632 (2025). Google Scholar Zhou, J., Müller, H., Holzinger, A. & Chen, F. Ethical ChatGPT: Concerns, challenges, and commandments. Electronics 13, 3417 (2024). Google Scholar Prosperi, M. et al. Causal inference and counterfactual prediction in machine learning for actionable healthcare. Nat. Mach. Intell. 2, 369–375 (2020). Google Scholar Thiese, M. S., Arnold, Z. C. & Walker, S. D. The misuse and abuse of statistics in biomedical research. Biochemia Med. 25, 5–11 (2015). Google Scholar Majnarić, L. T., Babič, F., O’Sullivan, S. & Holzinger, A. AI and big data in healthcare: Towards a more comprehensive research framework for multimorbidity. J. Clin. Med. 10, 766 (2021). Google Scholar Mahajan, S. The executioner paradox: understanding self-referential dilemma in computational systems. AI Soc. 40, 1939–1946 (2025). Google Scholar Mathews, S. C. et al. Digital health: a path to validation. npj Digital Med. 2, 38 (2019). Google Scholar Mueller, H., Mayrhofer, M. T., van Veen, E.-B. & Holzinger, A. The ten commandments of ethical medical AI. IEEE Computer 54, 119–123 (2021). Google Scholar Pairon, A., Philips, H. & Verhoeven, V. A scoping review on the use and usefulness of online symptom checkers and triage systems: how to proceed?. Front. Med. 9, 1040926 (2023). Google Scholar Wallace, W. et al. The diagnostic and triage accuracy of digital and online symptom checker tools: a systematic review. npj Digital Med. 5, 118 (2022). Google Scholar Pickard, M. D., Roster, C. A. & Chen, Y. Revealing sensitive information in personal interviews: Is self-disclosure easier with humans or avatars and under what conditions?. Comput. Hum. Behav. 65, 23–30 (2016). Google Scholar Bloice, M., Simonic, K.-M. & Holzinger, A. Casebook: a virtual patient iPad application for teaching decision-making through the use of electronic health records. BMC Med. Inform. Decis. Mak. 14, 1–9 (2014). Google Scholar De Togni, G., Erikainen, S., Chan, S. & Cunningham-Burley, S. What makes AI ‘intelligent’ and ‘caring’? Exploring affect and relationality across three sites of intelligence and care. Soc. Sci. Med. 277, 113874 (2021). Google Scholar Holzinger, A. & Mueller, H. Toward human-AI interfaces to support explainability and causability in medical AI. IEEE Computer 54, 78–86 (2021). Google Scholar Topol, E. J. High-performance medicine: the convergence of human and artificial intelligence. Nat. Med. 25, 44–56 (2019). Google Scholar Loveys, K., Sebaratnam, G., Sagar, M. & Broadbent, E. The effect of design features on relationship quality with embodied conversational agents: a systematic review. Int. J. Soc. Robot. 12, 1293–1312 (2020). Google Scholar Holzinger, A. et al. Personas for artificial intelligence (AI) an open source toolbox. IEEE Access 10, 23732–23747 (2022). Google Scholar Cabitza, F., Campagner, A. & Balsano, C. Bridging the “last mile” gap between AI implementation and operation: “data awareness” that matters. Ann. Transl. Med. 8, 501 (2020). Google Scholar Zuchowski, L. C., Zuchowski, M. L. & Nagel, E. A trust based framework for the envelopment of medical AI. npj Digital Med. 7, 230 (2024). Google Scholar Arnold, M. H. Teasing out artificial intelligence in medicine: an ethical critique of artificial intelligence and machine learning in medicine. J. Bioethical Inq. 18, 121–139 (2021). Google Scholar Longoni, C., Bonezzi, A. & Morewedge, C. K. Resistance to medical artificial intelligence. J. Consum. Res. 46, 629–650 (2019). Google Scholar Nag, P. K., Bhagat, A. & Priya, R. V. Expanding AI’s role in healthcare applications: a systematic review of emotional and cognitive analysis techniques. IEEE Access 13, 69129–69160 (2025). Google Scholar Okolo, C. T. Optimizing human-centered AI for healthcare in the Global South. Patterns 3, 100421 (2022). Google Scholar Rao, V. M. et al. Multimodal generative AI for medical image interpretation. Nature 639, 888–896 (2025). Google Scholar Download references This work was supported by the Austrian Science Fund (FWF) under grant 10.55776/P-32554 “Explainable Artificial Intelligence”. Machine Learning and Information Science Group, Diagnostic and Research Center for Molecular BioMedicine, Medical University of Graz, Graz, Austria Heimo Müller & Andreas Holzinger Human Machine Mind Corporation, Graz, Austria Heimo Müller The New York Academy of Medicine, New York, NY, USA Vimla L. Patel Department of Biomedical Informatics, Vagelos College of Physicians and Surgeons, Columbia University, New York, NY, USA Vimla L. Patel & Edward H. Shortliffe Schroeder Arthritis Institute and Krembil Research Institute, University Health Network; Departments of Medical Biophysics and Computer Science, and Faculty of Dentistry, University of Toronto; Institute of Neuroimmunology, Slovak Academy of Sciences, Bratislava, Slovakia; School of Digital Public Health, Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE, Toronto, ON, Canada Igor Jurisica Human-Centered AI Lab, FTEC, Department for Ecosystem Management, Climate and Biodiversity, University of Natural Resources and Life Sciences (BOKU), Vienna, Austria Andreas Holzinger Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar Search author on:PubMed Google Scholar A.H. and H.M. designed the survey poll. H.M. performed data analysis and prepared figures. A.H. and H.M. wrote the first draft. V.L.P., E.H.S., and I.J. contributed to conceptual framing and manuscript revision. All authors reviewed and approved the final version. Correspondence to Andreas Holzinger. The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions Müller, H., Patel, V.L., Shortliffe, E.H. et al. Seven deadly sins in artificial intelligence for digital medicine. npj Digit. Med. (2026). https://doi.org/10.1038/s41746-026-02607-4 Download citation Received: 08 November 2025 Accepted: 26 March 2026 Published: 15 April 2026 DOI: https://doi.org/10.1038/s41746-026-02607-4 Anyone you share the following link with will be able to read this content: Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative Advertisement npj Digital Medicine (npj Digit. Med.) ISSN 2398-6352 (online) © 2026 Springer Nature Limited Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

The two companies announced an expanded partnership at a Cadence conference in Santa Clara on Wednesday. The goal: make robot training data more accurate so physical AI systems reach real-world deployment faster. Cadence Design Systems and Nvidia have announced an expanded partnership aimed at closing one of robotics’ most persistent problems: the gap between how robots learn inside computer simulations and how they actually perform in the physical world. The collaboration, unveiled by the CEOs of both companies at a Cadence conference in Santa Clara, California, integrates Cadence’s high-fidelity physics simulation engines with Nvidia’s AI training platforms, including its Isaac open-source simulation libraries and Cosmos open-world models. Cadence is best known as one of the dominant suppliers of software used to design advanced computing chips. But the company also makes physics engines that model how real-world materials interact, how metals deform, how fluids flow, how surfaces make contact. These simulations are used in aerospace, automotive, and semiconductor design, but are now being applied to a new problem: generating the training data that robot AI systems need to learn how to handle objects and navigate physical environments. Training robots in simulation is faster and cheaper than doing so in the real world, but the training data is only as useful as the physics engine is accurate. “The more accurate the generated training data is, the better the model will be,” Cadence CEO Anirudh Devgan said at the Santa Clara conference. Nvidia CEO Jensen Huang described the scope of the collaboration directly: “We’re working with you across the board on robotic systems.” The combined stack will link Cadence’s multiphysics simulation with Nvidia’s model training pipelines and deploy the results on Nvidia’s Jetson robotics and edge AI hardware. The output is a workflow that runs from world-model training through physics simulation to real-world deployment feedback, coordinated by AI agents throughout the lifecycle. The announcement is part of a broader pattern of Nvidia building deep simulation partnerships across industrial engineering. The company has separately announced partnerships with Siemens and Dassault Systèmes to build industrial AI platforms and virtual twins. For Cadence, the robotics application represents a significant expansion of its simulation software into the AI infrastructure layer at a moment when demand for accurate robot training data is growing rapidly. I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, (show all) I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, and change. I write about startups, venture capital, digital policy, and Europe ecosystem, with an eye on the larger story beneath them: who gets to build the future, who profits from it, and how Europe is learning to speak in a louder voice of its own. Before moving into senior editorial leadership, I've built my career for over +10 years across journalism, storytelling, content strategy, SEO, and digital publishing, with experience in SaaS, hospitality, art, and culture. Get the most important tech news in your inbox each week. The heart of tech A Tekpon Company Copyright © 2006—2026, Cogneve, INC. Made with <3 in Amsterdam.

La startup américaine Figure AI vient de franchir une étape majeure dans l’histoire de la robotique humanoïde. Avec son nouveau modèle d’intelligence artificielle Helix 02, l’entreprise parvient à unifier locomotion, manipulation et équilibre au sein d’un même système neuronal, ouvrant ainsi la voie à des robots capables d’évoluer naturellement dans des environnements complexes et changeants. Jusqu’ici, les robots devaient souvent alterner entre déplacement et manipulation, au prix de mouvements saccadés et peu naturels. Helix 02 rompt avec cette approche fragmentée grâce à une architecture de type Vision-Language-Action pilotée par un réseau unique. Tous les capteurs — vision, toucher, perception du mouvement — sont ainsi connectés directement aux actionneurs, permettant au robot d’agir de manière fluide et sans interruption entre chaque action.  Le système repose sur trois niveaux complémentaires. Le niveau supérieur gère la compréhension du langage et des objectifs, tandis qu’un second niveau traduit la perception en gestes coordonnés. La grande nouveauté réside dans un troisième niveau fonctionnant à très haute fréquence, chargé de l’équilibre et de la stabilité, entraîné à partir de milliers d’heures de données sur le mouvement humain. Résultat : une motricité plus naturelle et une adaptation en temps réel aux contraintes physiques. Pour illustrer ses avancées, Figure AI a montré son robot effectuer seul une tâche domestique complexe : vider et remplir un lave-vaisselle dans une cuisine. Pendant plusieurs minutes, la machine a enchaîné des dizaines d’actions sans interruption, manipulant des objets fragiles, se déplaçant avec précision et coordonnant ses deux bras dans un espace contraint. Une performance rarement atteinte jusqu’ici par un robot humanoïde autonome. Helix 02 ouvre également la porte à des manipulations fines, comme dévisser un bouchon, extraire un comprimé ou trier de petits composants métalliques. En combinant vision rapprochée, capteurs tactiles et contrôle corporel global, Figure AI pose les bases d’une autonomie polyvalente, capable de s’adapter aux situations imprévues du monde réel. Si cette technologie reste pour l’instant cantonnée au laboratoire, elle marque aussi un tournant décisif dans la quête du robot humanoïde utile au quotidien. Helix 02 esquisse ainsi un futur où assistance domestique, logistique et industrie pourraient s’appuyer sur des machines réellement capables d’interagir avec leur environnement de manière fluide et intelligente. SOURCEGeneration-nt Signaler une erreur dans le texte Merci de nous avoir signalé l'erreur, nous allons corriger cela rapidement. Δ Nous nous réservons le droit de supprimer les commentaires qui ne respectent pas ces règles OpenAI voit partir deux dirigeants de plus au moment où l’entreprise démonte déjà une partie de son organisation... Chaque jour nous dénichons pour vous des promos sur les produits High-Tech pour vous faire économiser le plus d’argent possible. Voici... Anthropic a discrètement ajouté une page dédiée à la vérification d’identité pour son intelligence... YouTube modifie sa gestion des publicités pour les vidéos en direct (live) afin d’éviter de casser les séquences... Amazon ne se contente plus de tester Vega OS sur quelques produits : le groupe prépare désormais la transition de toute la gamme des Fire TV... Jeux Economie et entreprise Utilitaires Divertissement Economie et entreprise Météo Utilitaires Musique Jeux Jeux Drame Drame Action et aventure Drame Comédie Enfants / famille Enfants / famille Enfants / famille 18 Apr. 2026 • 8:00 18 Apr. 2026 • 7:00 17 Apr. 2026 • 22:40 17 Apr. 2026 • 19:44 Actualité High-Tech, Culture Geek et comparateur de prix Recherchez le meilleur prix des produits Hi-tech Recherchez des articles sur le site

A Figure mostrou mais uma impressionante demonstração do seu robot humanóide, a fazer tarefas domésticas. A Figure tem estado a trabalhar num sistema AI que permita aos robots humanóides fazer as tarefas que, para os humanos são “simples” mas para os robots são incrivelmente complexas. Agora, podemos ver o resultado desse trabalho com o modelo Helix 02 a permitir que um robot possa, de forma totalmente autónoma, tirar a loiça da máquina e arrumá-la nos sítios correctos, abrindo e fechando portas e gavetas, e também a colocar a loiça suja na máquina. Embora com ainda alguma lentidão face aos humanos, o robot demonstra movimentos surpreendentemente graciosos e fluidos, e sem as demoras de sistemas demonstrados no passado. Conta até com alguns pontos de destaque, como aos 2:50, quando após ter aberto uma gaveta, o robot faz um “toque de anca” para a fechar sem usar as mãos, ou aos 3:20, quando para fechar a porta da máquina de lavar começa por levantá-la com o pé – tal como a maioria dos humanos fará. Long‑Horizon Loco-Manipulation Helix 02 performs continuous, multi‑minute tasks that demand the full integration of locomotion, dexterity, and sensing pic.twitter.com/tweFSUMj5a — Figure (@Figure_robot) January 27, 2026 Como é habitual, não demoraram a surgir acusações de que esta demonstração terá sido feita com um humano a controlar remotamente o robot, mas o fundador da Figure, Brett Adcock (o mesmo que não acredita que as empresas chinesas estejam a produzir e vender centenas/milhares de robots humanóides) assegura que tudo foi deito de forma realmente autónoma graças ao modelo Helix 02. Tal como se previa, o ano de 2026 vai ser extremamente interessante a nível da evolução dos robots humanóides. O seu endereço de email não será publicado. Campos obrigatórios marcados com * Comentário * Nome * Email * Site Δ

AUSTIN, Texas (IT-Times) - Es wird nicht langweilig um den US-amerikanischen Elektrofahrzeug- und Batterieproduzenten Tesla. In Zukunft soll auch ein humanoider Roboter eine größere Rolle spielen. Es gibt mit Optimus Gen 2 eine neue Version. Tesla Inc. (Nasdaq: TSLA) präsentierte am 12. Dezember 2023 per Video den Optimus Gen 2 Roboter, auch als Tesla Bot bekannt, mit verbesserten Händen und schlankerer Silouette. Das Video zeigt Verbesserungen an seinem humanoiden Roboter-Prototyp. Zwei der Maschinen tanzen unter blinkenden Lichtern zu elektronischer Musik. Die neue Version des Roboters kann in die Hocke gehen, ohne umzufallen. Die als Optimus-Bot bezeichnete Maschine, die den Menschen nachahmt, ist Teil von Teslas Unterfangen zur Entwicklung künstlicher Intelligenz (KI) und nutzt ein trainiertes neuronales Netzwerk, um grundlegende Aufgaben auszuführen. Bild: Tesla - Optimus Gen 2 Das Unternehmen gab an, dass der Roboter im Vergleich zu seinem Vorgängermodell 30 Prozent schneller laufen könne, 10 kg leichter sei und über ein verbessertes Gleichgewicht und bessere Handbewegungen verfüge. Der Clip zeigt, wie der Roboter seine Finger beugt und ein Ei kocht, um Fortschritte bei der „empfindlichen Objektmanipulation“ zu demonstrieren. Meldung gespeichert unter: Elektroauto, Roboter (Robotik), Elektromobilität, Elon Musk, Automobile, Tesla, Hardware, Software © IT-Times 2026. Alle Rechte vorbehalten. Erhalten Sie einen Wissensvorsprung! Abonnieren Sie unseren 2x wöchentlich erscheinenden Newsletter mit den relevantesten Business-Nachrichten der Woche. Weitere Möglichkeiten, um auf dem Laufenden zu bleiben, haben wir in einer Übersicht für Sie zusammen gestellt. Sie haben die Möglichkeit, mit unserem Webmaster-Nachrichten-Tool die Nachrichten von IT-Times.de kostenlos auf Ihrer Internetseite einzubauen. Zugeschnitten auf Ihre Branche bzw. Ihr Interesse. Weitere Möglichkeiten, um auf dem Laufenden zu bleiben, haben wir in einer Übersicht für Sie zusammen gestellt. Erhalten Sie einen Wissensvorsprung! Abonnieren Sie unseren 2x wöchentlich erscheinenden Newsletter mit den relevantesten Business-Nachrichten der Woche. Sie haben die Möglichkeit, mit unserem Webmaster-Nachrichten-Tool die Nachrichten von IT-Times.de kostenlos auf Ihrer Internetseite einzubauen. Zugeschnitten auf Ihre Branche bzw. Ihr Interesse.

Tesla, la marque qui a redéfini le marché des véhicules électriques, continue d’innover avec sa dernière création : la station de recharge Tesla Optimus. Ce développement suscite beaucoup d’enthousiasme à travers le monde automobile, alors que la demande en infrastructures de recharge s’intensifie. Le concept derrière la station de recharge Tesla Optimus s’aligne avec la vision de durabilité et d’efficacité énergétique de l’entreprise. Avec une capacité de chargement rapide inédite, elle est conçue pour répondre aux besoins croissants des utilisateurs de véhicules électriques (VE). Les stations Optimus visent à réduire le temps de recharge tout en augmentant la capacité de service simultané pour plusieurs véhicules. Tesla Optimus Charging Station pic.twitter.com/9SH5eqAfs5 Un des points forts des stations Tesla Optimus est sans aucun doute l’intégration de technologies de pointe. Chaque station comprend un système de gestion intelligent permettant d’optimiser le flux énergétique selon l’état de la demande, garantissant ainsi une expérience utilisateur fluide. De plus, grâce à l’application Tesla, les utilisateurs peuvent suivre en temps réel le statut de leur recharge et planifier leurs trajets en conséquence. Alors que les voitures électriques gagnent du terrain en tant que choix principal pour de nombreux consommateurs, l’infrastructure de recharge est plus que jamais sous pression. L’introduction des stations Optimus contribue non seulement à soulager cette pression, mais aussi à renforcer le réseau électrique global. En travaillant en partenariat avec des fournisseurs d’énergie renouvelable, Tesla assure que ses stations sont alimentées de manière durable. L’arrivée de ces stations innovantes est particulièrement favorable pour le marché européen, où la transition vers des solutions plus écologiques est prioritaire. L’expansion des stations Tesla pourrait accélérer l’adoption des VE en Europe, offrant aux utilisateurs des solutions pratiques et écologiques. Avec la station de recharge Tesla Optimus, Tesla continue de mener la révolution électrique mondiale. Ces infrastructures de recharge avancées sont plus qu’un simple ajout à l’écosystème de Tesla ; elles sont le reflet d’une stratégie audacieuse qui vise à rendre le transport électrique accessible, fiable et durable. Alors que le monde avance vers un avenir plus propre, Tesla semble bien positionnée pour être à l’avant-garde de ce mouvement transformateur. Fondateur de Tesla Mag | Analyste Stratégique Mobilité & IA Observateur privilégié de l'écosystème Tesla depuis 2013, il décrypte les ruptures technologiques d'Elon Musk avec une rigueur d'expert. Spécialiste des infrastructures IRVE et des marchés Tesla Energy (100 GW), il analyse l'impact de l'IA sur la conduite autonome (FSD) et l'industrie automobile mondiale. Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec * Commentaire * Nom * E-mail * Enregistrer mon nom, mon e-mail et mon site dans le navigateur pour mon prochain commentaire. Leave this field empty Leave this field empty Δ Tesla Mag est un média dédié aux véhicules électriques Premium. Nous publions des actualités, guides d'achat et Bons plans pour vous permettre de découvrir l'univers VE. © 2026 Tesla Mag - Tous droits réservés. La reproduction intégrale ou partielle des contenus, articles, et images sans autorisation explicite est interdite.

Релоцировались? Теперь вы можете комментировать без верификации аккаунта. Компания временно приостановила производство гуманоидных роботов Optimus из-за проблем с конструкцией рук и предплечий. Компания временно приостановила производство гуманоидных роботов Optimus из-за проблем с конструкцией рук и предплечий. Компания временно приостановила производство гуманоидных роботов Optimus из-за проблем с конструкцией рук и предплечий. По данным The Information, инженеры компании не смогли добиться необходимой ловкости движений, близкой к человеческой. Именно эта часть конструкции оказалась наиболее сложной для реализации. По данным источников, на предприятиях Tesla уже накопились десятки корпусов роботов без рук и предплечий, а сроки завершения их сборки остаются неопределенными. Первоначально Илон Маск планировал выпустить 5000 единиц Optimus до конца 2025 года. Однако из-за выявленных проблем компания пересмотрела цели: теперь речь идет максимум о 2000 роботах, и даже этот показатель может быть отложен. Маск признал наличие трудностей, отметив, что создание рук с человеческой степенью ловкости стало самым сложным этапом проекта. Сроки возобновления производства он не назвал, но подчеркнул, что проект будет продолжен. Несмотря на задержки, Tesla продолжает показывать развитие технологии. Недавно Маск опубликовал видео, на котором робот Optimus выполняет приемы кунг-фу, а также ролик, где он повторяет движения актера Джареда Лето на премьере фильма Tron: Ares. Tried to start a fight at the Tron: Ares premiere pic.twitter.com/TvWCOaXIlN Робот Optimus был впервые представлен в 2021 году как универсальный гуманоид, способный выполнять рутинные и опасные для человека задачи. Маск заявлял, что в будущем производство таких машин может стать для Tesla даже более прибыльным направлением, чем электромобили. Как помочь, если вы в Польше Хотите сообщить важную новость? Пишите в Telegram-бот Главные события и полезные ссылки в нашем Telegram-канале Релоцировались? Теперь вы можете комментировать без верификации аккаунта. (руки) растут из одного места??? Пользователь отредактировал комментарий 14 октября 2025, 15:45

Напомним, что гуманоидного робота Optimus представили в октябре этого года на презентации We, Robot. В перспективе цена одного такого робота составляет 20—30 тысяч долларов. Также в октябре был представлен Cybercab — беспилотное такси, в котором нет ни руля, ни педалей. Сейчас новинки Tesla смогла протестировать Ким Кардашьян, одна из самых популярных инфлюенсеров в мире. Ким удалось провести время и с классической, и с позолоченной версией Optimus (последняя была выпущена в единственном экземпляре). В видеороликах блогер сыграла с роботом в «‎камень-ножницы-бумагу», попросила Optimus показать «‎сердечко» и прокомментировала его танец. meet my new friend @Teslapic.twitter.com/C34OvPA2dY — Kim Kardashian (@KimKardashian) November 18, 2024 Позолоченный Optimus находился в салоне Cybercab. Ким назвала беспилотный автомобиль Tesla, производство которого должно начаться в 2027 году, «невероятным» и «безумным». Отметим, что эти видео не являются спонсорскими, но один из постов Ким ретвитнул Илон Маск. Optimus is here to take mw for a ride in the Cybercab pic.twitter.com/gxOSbsY3vv — Kim Kardashian (@KimKardashian) November 19, 2024 Есть о чем рассказать? Пишите в наш телеграм-бот. Это анонимно и быстро

Tesla, la société pionnière dans le domaine des véhicules électriques, continue de faire grand bruit avec ses projets d’expansion ambitieux. Dernièrement, l’attention s’est portée sur l’annonce faite lors de la réunion annuelle des actionnaires, selon laquelle la construction d’une usine de production massive à Giga Texas est en cours. Tesla a dévoilé son intention de créer une installation capable de produire 10 millions d’unités par an. Ce projet titanesque non seulement solidifie la position de Tesla en tant que leader mondial dans le domaine des technologies électriques, mais promet également de transformer le paysage industriel. Initialement célèbre pour sa capacité à produire des véhicules électriques de pointe, Giga Texas est appelé à durer sous cette nouvelle expansion. Cet emplacement stratégique non seulement élargit le champ d’action de Tesla, mais vise également à stimuler l’économie locale en créant de nombreux emplois. La mise en place de l’usine Optimus intègre également des innovations technologiques avancées. Cette nouvelle unité de production promet de réduire considérablement les délais tout en maintenant une qualité de produit exceptionnelle. Grâce à son processus de fabrication de pointe, Tesla Optimus pourrait redéfinir la façon dont les véhicules électriques sont produits dans un avenir proche. Alors que l’industrie automobile continue de se diriger vers des pratiques plus durables, l’usine Optimus devrait jouer un rôle clé dans la réduction de l’empreinte carbone mondiale. Les efforts de Tesla alignent la production de masse avec ses engagements en faveur de la durabilité. En outre, cette expansion devrait avoir un impact économique positif, facilitant la croissance des opportunités d’emploi et stimulant le développement économique régional. Avec cette initiative révolutionnaire, Tesla démontre son engagement à rester à la pointe de l’innovation technologique. La réalisation du projet Optimus à Giga Texas pourrait bien annoncer une nouvelle ère dans le secteur des véhicules électriques, influençant la direction future du transport durable à l’échelle mondiale. En conclusion, le projet d’usine Optimus de Tesla à Giga Texas représente bien plus qu’une simple expansion industrielle. Il s’agit d’une étape significative vers l’avenir des véhicules électriques et de notre planète. L’engagement de Tesla pour le progrès technologique et la durabilité est plus fort que jamais, et les yeux du monde entier sont tournés vers le Texas pour voir comment cette histoire pionnière se déroulera. Fondateur de Tesla Mag | Analyste Stratégique Mobilité & IA Observateur privilégié de l'écosystème Tesla depuis 2013, il décrypte les ruptures technologiques d'Elon Musk avec une rigueur d'expert. Spécialiste des infrastructures IRVE et des marchés Tesla Energy (100 GW), il analyse l'impact de l'IA sur la conduite autonome (FSD) et l'industrie automobile mondiale. Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec * Commentaire * Nom * E-mail * Enregistrer mon nom, mon e-mail et mon site dans le navigateur pour mon prochain commentaire. Leave this field empty Leave this field empty Δ Tesla Mag est un média dédié aux véhicules électriques Premium. Nous publions des actualités, guides d'achat et Bons plans pour vous permettre de découvrir l'univers VE. © 2026 Tesla Mag - Tous droits réservés. La reproduction intégrale ou partielle des contenus, articles, et images sans autorisation explicite est interdite.

Робот Tesla Optimus третьего поколения приступает к работе, о чем говорится на официальной страничке разработчиков. Заведение на бульваре Санта-Моника открылось летом 2025 года и быстро привлекло внимание благодаря необычному формату: ретро-футуристическое автокафе с зарядной станцией и гуманоидным роботом, раздающим попкорн. Версия второго поколения, получившая неофициальное прозвище Poptimus, стала одной из главных «фишек» площадки. Однако к декабрю 2025 года робот исчез — это было частью подготовки к более серьёзному обновлению. Ранее Илон Маск подтвердил, что в 2026 году Optimus вернётся уже в новом качестве. Вместо демонстрационных функций он сможет выполнять роль курьера, доставляя заказы прямо к автомобилям на станциях Supercharger. Ключевым обновлением стал переход к третьему поколению. Новый Optimus получил значительно улучшенную механику рук — около 50 приводов и 22 степени свободы на каждую, а также мощную вычислительную платформу с чипом Tesla AI5 и голосовым управлением на базе Grok. Маск уже назвал его самым продвинутым роботом в мире. Параллельно компания перестраивает производство: линии во Фримонте планируют переоборудовать под выпуск Optimus, что подчёркивает смену приоритетов Tesla в сторону робототехники. © iXBT

Home » Artificial intelligence » Tesla Optimus Talk Konstantinos Laskaris, Tesla Lead Director of Optimus, at the ETH Robotics Club INSPIRE Talk in Switzerland. Konstantinos presented Optimus 2.5 and the work behind it and its predecessors. Here are some highlights from the talk. The sim-to-real gap is propaganda. “It’s not a gap if you haven’t tried to model your robot properly.” Hardware matters more than most people think. If you can’t replicate human motion on hardware, no amount of real-world data will save you. Tendon-driven hands are the way. Anything with motors physically cannot reproduce human muscle force density. Not a preference: a physics constraint. On physics engines: don’t constrain yourself to what exists. His challenge to the community: do you even understand how physics works? Go build your own simulation. Reproduce the fidelity YOU need. Optimus V3 is coming soon. It won’t be sold to the public. It won’t go to factories. First customer? Tesla itself. They’re building a Bot Academy. A secure environment where robots learn tasks from scratch. Example: hold and operate a drill. Optimus V3 is coming soon. It won’t be sold to the public. It won’t go to factories. First customer? Tesla itself. They’re building a “Bot Academy”: a secure environment where robots learn tasks from scratch. Example: hold and operate a drill. — odesha (@oskrt_dvs) April 3, 2026 Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology. Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels. A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements. “It won’t be sold to the public. It won’t go to factories.” The first part seems obvious and what they’ve said for a long time. The second is just confusing. What’s the point in mass producing them if they don’t use them throughout the Musk enterprises and with suppliers? They don’t need tens of thousands of them for a bot academy. Maybe this part is just about the first few months of slow production. By “factories” he was talking about selling them commercially, not themselves. Musk will using them in his factories, doing real work, replacing real people, this year. I didn’t know they wont even sell V3. I knew the 1st year or so would go to Tesla themselves, and perhaps some to other Musk ventures. But I thought at some point next year, anyone could buy one. I guess customers will have to wait for V4, which will have a AI5 chip, and I’d bet begin to be built late next year. Certainly feels like they continue to fall behind, But I think that’s because Elon revealed WAY too much during his retarded AI day, and everyone from China was drooling with pen & paper in hand. So now he keeps it closer to the chest. But he just needs to show things off better, focus on capabilities, not hardware or software. “Tendon-driven hands are the way. Anything with motors physically cannot reproduce human muscle force density. Not a preference: a physics constraint.” Yes, some sort of variable torsion device that will allow the delicate grasping of a raw egg without breakage and yet also will allow heavy work such as tightening up bolts while holding heavy axle hubs, etc….. it gets around the difficulty of trying to match the dexterity of 6 million years of hominin/hominid evolution by using a different approach. https://www.nature.com/scitable/knowledge/library/overview-of-hominin-evolution-89010983 AI Robotics will advance as more practical tests are performed in “real world” situations. I can see this technology making great strides into automobile and “white goods” manufacturing, possibly into ship welding/construction as well. Comments are closed.

Generations in Dialogue: Bridging Perspectives in AI is a podcast from AAAI featuring thought-provoking discussions between AI experts, practitioners, and enthusiasts from different age groups and backgrounds. Each episode delves into how generational experiences shape views on AI, exploring the challenges, opportunities, and ethical considerations that come with the advancement of this transformative technology. In the fourth episode of this new series from AAAI, host Ella Lan chats to Professor Marynel Vázquez about what inspired her research direction, how her perspective on human-robot interactions has changed over time, robots navigating the social world, potential for using robots in education, modeling interactions as graphs, addressing misunderstandings with regards to robots in society, getting input from target users, the challenge of recognising when errors happen, making robots that adapt, and more. Marynel Vázquez is a computer scientist and roboticist whose research focuses on Human-Robot Interaction (HRI), particularly in multi-party settings. She studies social group dynamics—such as spatial behavior and social influence—in HRI, and develops perception and decision-making algorithms that enable autonomous, socially aware robot behavior. A central theme in her work is modeling interactions as graphs, allowing robots to reason about individuals, relationships, and groups simultaneously. Her interdisciplinary approach combines computer science, behavioral science, and design, and she enjoys building new robotic systems and research infrastructure to bring theoretical ideas into real-world practice. Ella Lan, a member of the AAAI Student Committee, is the host of “Generations in Dialogue: Bridging Perspectives in AI.” She is passionate about bringing together voices across career stages to explore the evolving landscape of artificial intelligence. Ella is a student at Stanford University tentatively studying Computer Science and Psychology, and she enjoys creating spaces where technical innovation intersects with ethical reflection, human values, and societal impact. Her interests span education, healthcare, and AI ethics, with a focus on building inclusive, interdisciplinary conversations that shape the future of responsible AI.

Nachrichten, Termine, Experten d Instanz: Teilen Teilen: 14.04.2026 16:03 Positions were in high demand: 20 international teams – more than ever before – will face one of the world’s most demanding performance tests for military robotics at the European Land Robot Trial (ELROB) in mid-June. The impressive venue for this four-day major event is the Thun military training area, which the Swiss Federal Office for Defense Procurement (armasuisse) is providing as host in collaboration with the Swiss Army. Here, the participants will compete with their robotic systems in several disciplines, whose realistic scenarios have been developed by a team led by Dr Frank E. Schneider from the Fraunhofer Institute for Communication, Information Processing and Ergonomics FKIE. “This year’s field of participants is particularly exciting,” says Schneider, looking at the registration list. “An interesting mix of established teams we already know from previous events and those taking part in ELROB for the very first time.” Among the latter, the deputy head of the FKIE’s “Cognitive Mobile Systems” (CMS) department cites Team Łukasiewicz-PIAP from Poland and the two German teams GAP and FENRIDE. Further participants are travelling to Thun from the Netherlands, the Czech Republic, Austria and Switzerland, whilst two teams are coming to ELROB this year specifically from Canada. Highly realistic scenarios In the main disciplines of Reconnaissance, Transport (Mule) and Search & Rescue (SAR), the teams will put their Unmanned Ground Vehicles (UGVs) and Unmanned Aerial Vehicles (UAVs) through their paces over four days. Common to all scenarios is the high level of realism and the close alignment with the current needs of the armed forces. It is no coincidence that, against this backdrop, several teams have shown an interest in the "Mule" discipline, for example, which is "closely aligned with procurement", as Schneider explains: "The transport of personnel and equipment is an essential component of military operations. In hostile environments, however, this is a dangerous and demanding task, which is why UGVs are increasingly being deployed here." In Thun, their practical suitability can be tested and demonstrated to the fullest extent. The military training area is not only the oldest but, at around 6.5 square kilometres, also the largest of its kind in Switzerland. In its centre, a large tent city is being set up for the participants, where they can program, tinker with and fine-tune their robots and drones around the clock. For Schneider, who has been organising ELROB every two years with his team since 2006, the venue is familiar. Thanks to a trilateral R&D cooperation between Germany, Austria and Switzerland, the host country for the competition rotates every two years: “Thun was already the venue in 2012 and we are delighted to be back here now,” says Schneider. For ELROB host Dr Thomas Rothacher, Head of armasuisse Science and Technology and Deputy Chief of Defence, cross-border cooperation enables “a valuable exchange of experience and knowledge.” At the same time, the event offers “a unique opportunity to test robotics technologies in military operations and thereby strengthen security-related robotics research between industry, universities and national and international partners,” says Rothacher. Constantly redesigned scenarios The team is not revealing any details about the scenarios. Suffice it to say that this time there is no urban environment; there are no buildings or other structures to explore. The tasks require different approaches and solutions, which will, incidentally, be assessed by an international jury led by the renowned robotics expert Prof. Dr Henrik I. Christensen. “The demands on robotics are increasing rapidly,” says ELROB initiator Schneider. “And we are responding to this by constantly redesigned scenarios.” He is particularly pleased with the diverse field of participants from research, universities and industry: “This shows once again that ELROB more than lives up to its claim of bringing users, researchers and clients together.” European Land Robot Trial 15 to 19 June 2026 Thun Military Base, Switzerland Dr Frank E. Schneider, Cognitive Mobile Systems Department Fraunhofer Institute for Communication, Information Processing and Ergonomics FKIE frank.schneider@fkie.fraunhofer.de I Telephone: +49 228 9435481 https://www.fkie.fraunhofer.de/en/press-releases/2026-elrob.html http://www.fkie.fraunhofer.de/elrobhttp://www.elrob.org <What is the current state of robotics? At ELROB, teams are testing their unmanned ground and aerial ...Quelle: Fabian VoglCopyright: Fraunhofer FKIE <A dedicated tent city is being set up at the Thun military training area, where the teams will prepa ...Quelle: Fabian VoglCopyright: Fraunhofer FKIE Merkmale dieser Pressemitteilung: Journalisten, Wirtschaftsvertreter, Wissenschaftler Informationstechnik überregional Forschungs- / Wissenstransfer, Wettbewerbe / Auszeichnungen Englisch <What is the current state of robotics? At ELROB, teams are testing their unmanned ground and aerial ...Quelle: Fabian VoglCopyright: Fraunhofer FKIE <A dedicated tent city is being set up at the Thun military training area, where the teams will prepa ...Quelle: Fabian VoglCopyright: Fraunhofer FKIE Zum Download Zum Download Suche in Pressemitteilungen Suche in Terminen Anfangsdatum Enddatum Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie. Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie). Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“. Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet). Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).

SHENZHEN, China, April 2, 2026 /PRNewswire/ — X Square Robot, an emerging leader in embodied AI and humanoid robotics, announced the successful conclusion of the world’s first Embodied AI Developers Conference (EAIDC 2026), the first global gathering dedicated specifically to developers building embodied AI systems. EAIDC 2026 marks one of the first large-scale industry gatherings dedicated to embodied AI, bringing together leading researchers, developers, and technology companies to accelerate the transition of intelligent systems from laboratory research to real-world applications. The event features live robotic demonstrations, a national-level hackathon, and discussions between academia and industry focused on deployment, commercialization, and ecosystem development. EAIDC 2026, the world’s first embodied intelligence developer competition, is designed to accelerate real-world innovation through hands-on collaboration and deployment-focused challenges. X Square Robot hosted EAIDC as part of a broader effort to contribute to the global embodied AI ecosystem and expand its role in shaping the future of intelligent systems. The competition also introduced a set of “three firsts” designed to bring embodied AI development closer to real-world conditions, including real-robot task execution, continuous system evaluation, and full end-to-end deployment workflows. Task design focused on four core capability areas — grasping and placement, language understanding, fine manipulation, and long-horizon decision-making — with participants completing challenges such as ring placement, instruction-based fruit sorting, cable plugging, and word spelling. A full-variable evaluation approach required teams to operate without preset parameters, using randomized real-world environments to test true adaptability and model robustness. The company is operating at a time of growing global momentum around humanoid robotics, as advances in physical AI and vision-language-action (VLA) models begin to unlock more complex, real-world capabilities. The company has raised approximately $280 million to date, with backing from investors including Alibaba, ByteDance, Meituan, HongShan (formerly Sequoia China), and other leading technology and venture firms. X Square Robot is focused on developing general-purpose humanoid systems capable of operating in dynamic, unstructured environments. Its technology centers on embodied foundation models designed to enable robots to perceive, adapt, and perform tasks across a range of real-world scenarios. The company has begun generating early revenue from deployments across sectors including education, hospitality, and elder care, and is exploring broader applications in household services through collaborations such as its partnership with 58.com. The company has also been active in advancing technical research and industry engagement, including participation in global AI forums and academic communities such as CVPR, reflecting its commitment to contributing to both the development and commercialization of embodied intelligence. These efforts reflect a broader industry shift toward applying AI-driven robotics to address labor shortages and operational challenges in both consumer and commercial environments. By hosting EAIDC 2026, X Square Robot is engaging with a growing global ecosystem of developers and industry stakeholders working to define the next generation of intelligent systems. The company views this as part of a broader strategy to expand its international presence and establish itself as a global innovator in embodied AI. For more information, visit https://x2robot.com/. About X Square Robot X Square Robot is a global innovator in embodied artificial intelligence, robotics, and autonomous systems. The company develops integrated software and hardware platforms that enable robots to perceive, reason, and act safely in complex environments. Partnering with leading universities and technology institutions, X Square Robot delivers scalable AI infrastructure and establishes open benchmarks to drive worldwide progress in intelligent robotics. ​ Your email address will not be published. Required fields are marked * Comment * Name * Email * Website Save my name, email, and website in this browser for the next time I comment. Copyright © 2024 Money Compass Media (M) Sdn Bhd. All Rights Reserved Login to your account below Remember Me Please enter your username or email address to reset your password. Copyright © 2024 Money Compass Media (M) Sdn Bhd. All Rights Reserved

From daily news and career tips to monthly insights on AI, sustainability, software, and more—pick what matters and get it in your inbox. Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. Follow Us On Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. System hits 310 units/hour, 19–20 sec cycles, 99 percent success; integrated in 36 hours, producing about 3,000 units per shift. Chinese robotics player Agibot has revealed that it has deployed its humanoid robots in a live manufacturing facility. The rollout of Agibot G2 robots at the Shagahai-based electronics manufacturer Longreacher Technology’s facility marks a step toward real-world industrial adoption of embodied AI systems. The Agibot G2 units are now operating at multimedia-integrated testing stations, where they handle precise loading and unloading tasks. According to Agibot, the deployment highlights the growing role of humanoid robots in manufacturing environments, demonstrating their ability to perform repetitive, high-accuracy operations alongside existing production workflows. In March, Agibot announced the rollout of its 10,000th humanoid robot, marking a major milestone in embodied AI industrialization. At Longcheer Technology’s production facilities, Agibot G2 robots are currently deployed at its multimedia-integrated testing stations. The robots are responsible for executing precision loading and unloading tasks. These operations require a high degree of accuracy, consistency, and coordination, demonstrating the maturity of Agibot’s hardware and software systems in handling repetitive industrial processes. The testing stations combine multiple functional modules, requiring seamless interaction between perception, motion planning, and manipulation. According to Agibot, G2 robots leverage multi-modal sensing capabilities, including visual perception and spatial awareness, to accurately identify objects and execute task sequences. Their ability to operate continuously within structured production workflows highlights their readiness for industrial adoption. According to Agibot, the deployment is broader in strategy, building a full-stack ecosystem for embodied intelligence. The company integrates robot hardware, AI models, and large-scale data infrastructure to enable continuous learning and improvement. Through this approach, robots are not limited to predefined instructions but can adapt to variations in tasks and environments over time. “This project shows that embodied AI is no longer experimental. It is a practical, production‑ready capability that can operate reliably in real industrial environments and deliver measurable economic value,” said Maoqing Yao, Partner, Senior Vice President, and President of the Embodied Business Unit at Agibot, in a statement. The system is capable of operating in high-precision manufacturing tasks, navigating complex factory layouts, placing devices into testing fixtures with millimeter-level accuracy, and sorting finished or defective units accordingly. Unlike conventional industrial automation, the system requires no custom tooling and supports mixed-model production, enabling faster changeovers and significantly reducing downtime. Agibot claims the deployment has demonstrated strong, quantifiable performance across key industrial metrics, including throughput of up to 310 units per hour, cycle times of approximately 19–20 seconds per operation, and a success rate exceeding 99 percent in continuous operation. Production line integration was completed within 36 hours, with output reaching approximately 3,000 units per shift. The system supports 24/7 autonomous operation with minimal human intervention, achieving over 140 hours of cumulative continuous operation while maintaining downtime loss below 4 percent. A single Agibot G2 robot can replace multiple manual processes while maintaining consistent output, enabling manufacturers to balance efficiency, cost, and flexibility in a unified system. The system’s performance is driven by Agibot’s embodied AI approach, allowing robots to be deployed quickly, adapt to changing production conditions, and operate reliably in high-speed manufacturing environments. By combining simulation-based validation, reinforcement learning, and on-device intelligence, the system minimizes setup time, reduces the need for manual adjustments, and ensures stable performance in continuous production. With multiple units already in operation, Agibot plans to expand deployment to 100 robots by Q3 2026, accelerating adoption across industries including automotive, semiconductors, and energy. Experts suggest the developments reflect a broader shift in manufacturing from rigid, hardware-defined automation toward flexible, software-driven intelligent systems powered by embodied AI. Jijo is an automotive and business journalist based in India. Armed with a BA in History (Honors) from St. Stephen's College, Delhi University, and a PG diploma in Journalism from the Indian Institute of Mass Communication, Delhi, he has worked for news agencies, national newspapers, and automotive magazines. In his spare time, he likes to go off-roading, engage in political discourse, travel, and teach languages. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Premium Follow

Add the following snippet to your HTML:<iframe frameborder='0' height='385' scrolling='no' src='https://www.hackster.io/HiwonderRobot/from-chatbots-to-bodies-embodied-ai-on-rosorin-pro-8e3215/embed' width='350'></iframe> Stop restricting AI to a screen. Learn how ROSOrin Pro gives LLMs a physical presence to sense, move, and interact with the 3D world. 🤖🌍 Read up about this project on Stop restricting AI to a screen. Learn how ROSOrin Pro gives LLMs a physical presence to sense, move, and interact with the 3D world. 🤖🌍 The biggest hurdle in modern robotics isn't the AI—it’s the "hand-off" between software logic and hardware execution. This is why the ROSOrin Pro is designed to be OpenClaw Ready out of the box. OpenClaw is more than just a gripper standard; it is a unified framework that allows Large Language Models (LLMs) to communicate with physical actuators. By being "OpenClaw Ready," the ROSOrin Pro ensures that developers can skip the nightmare of low-level driver integration. Whether you are deploying a custom GPT-based agent or a local Llama 3 node, the platform provides a plug-and-play interface for complex 3D manipulation, allowing your AI to focus on the "Thinking" while the hardware handles the "Doing." Why does "Embodied AI" matter? A chatbot can describe how to pick up a cup, but it lacks a "nervous system" to feel the weight or see the steam. Embodied AI is about Spatial Grounding—linking digital tokens to physical coordinates. The ROSOrin Pro serves as the ultimate development sandbox for this evolution. Powered by the NVIDIA Jetson Orin Nano or Raspberry Pi 5, it provides the high-performance edge computing necessary to run multimodal models that perceive, reason, and act in real-time. It transforms a static AI into a mobile agent capable of navigating human environments. An intelligent agent is only as smart as its data. The ROSOrin Pro utilizes a sophisticated sensory suite to "ground" its intelligence: To ensure the AI’s "thoughts" translate into fluid motion, the ROSOrin Pro runs on ROS 2 Humble. This middleware acts as the robot's backbone, managing the high-speed communication between the AI vision nodes and the 6-DOF robotic arm. With built-in Inverse Kinematics (IK), the ROSOrin Pro can calculate complex trajectories on the fly. When the AI decides to "Move the red block to the bin," the ROS 2 stack autonomously plans the arm’s path, avoiding obstacles and maintaining balance, effectively acting as the robot’s "motor cortex." Embodied AI is a two-way street. Using the onboard AI Voice Interaction Module, the ROSOrin Pro creates a multimodal feedback loop. It doesn't just take orders; it interacts. If an object is out of reach or too heavy, the robot can communicate this back to the user or the LLM to refine the task. This level of "Reasoning-in-the-Loop" is what separates a smart car from a true autonomous assistant. The era of physical AI agents is just beginning, and we want you to lead the charge. We are excited to announce that our comprehensive OpenClaw gameplay tutorials for the ROSOrin Pro are launching soon! These upcoming guides will cover everything from local LLM deployment to 3D visual grasping algorithms. Don’t miss out on the next wave of Embodied AI: Follow Hiwonder on GitHub to access our open-source codebases and pre-configured ROS 2 images. Watch our Hackster Profile for free project guides and cutting-edge developer resources. Hackster.io, an Avnet Community © 2026

The ground vehicle ULTRA from Ukrainian company Overland Al allows operators to deploy multiple drones with no human present. Courtesy OVERLAND AI Stay Connected Patrick Tucker NATO is studying how to use ground and air robots to replace human soldiers in assaults, something Ukraine has been doing for more than a year. But that hasn’t stopped Russia’s continuous assault with its own, increasingly autonomous one-way attack drones. On Tuesday, Ukrainian President Volodymyr Zelensky made a social-media splash with a video describing a historic first from last July: a skirmish in which Russian troops surrendered to Ukrainian robots. “The future is already on the front line—and Ukraine is building it,” Zelenskyy said in the video, adding that Ukrainian robotics companies “have already carried out more than 22,000 missions on the front in just three months.” Still, the Ukrainian president offered far fewer details than did Ukraine’s 3rd Assault Brigade in its own July 2025 post. “Enemy fortifications were attacked” by first-person-view aerial drones and ground robots armed with explosives and made by Nazemnyi Robotychnyi Kompleks, the post said. “The next robot was already approaching the destroyed dugout when the enemy, in order to avoid being blown up, announced surrender. The occupiers who survived were taken to our lines by ‘birds’ [aerial drones] and, according to the regulations, taken prisoner.” “The operation was carried out without infantry and without losses on our side,” it said. “The occupiers surrendered to the ground robots of the Third Assault!” Ukraine’s ground-robot game advanced quickly in the following months, said Olena Kryzshanivska, a senior editor at the NATO Association of Canada who first relayed the news to English-language audiences. “Already…[by the] beginning of this year, we saw several documented cases when UGVs [unmanned ground vehicles] were used for strike missions. They were either delivering grenades [or] they were sometimes … attacking trenches, attacking Russian troops,” Kryzshanivska said in February during a podcast with CNAS adjunct senior fellow Sam Bendett. That sort of combined robotic fast maneuver is one of the ways Ukraine is forcing a reconsideration of decades of military doctrine, and NATO is taking notice. In February, its Allied Command Transformation announced the extension of a study on Force Lethality Enhancement to build out “a few practical force options and test them against realistic scenarios to see what works, and what it would take to use them on operations.” Another alliance effort to integrate ground robots, part of the multidomain Task Force X, is being led by Brig. Gen. Chris Gent, NATO deputy chief of staff transformation and integration. Venture capitalists are taking note as well. Eric Brock of Ondas Capital told Defense One in January that his firm is investing in “ground robots that are tailored towards defense and homeland security but also critical infrastructure protection in certain places.” Challenges The biggest constraint in using first-person-view drones is that an operator can generally fly just one at a time. But the drone can fly itself to waypoints, loiter in the air, and reconnect after brief communications interruptions. Ground robots, by contrast, need constant attention because navigation remains a technical challenge, John Hardie, of the Foundation for the Defense of Democracies, told reporters in February. And UGV operators must also stay in frequent contact with the operators of the aerial drones above. “My understanding is that they've experimented with autonomous navigation, but it’s especially difficult with [unmanned ground vehicles] for that to be reliable. So I don't think they're there yet,” Hardie said. Ukraine has also been hunting for alternatives to GPS, which is jammable. Since 2023, it has been experimenting with visual- and terrain-matching systems and other AI-powered ideas for long-range navigation, Hardie said. Russia, too, has carried out robotic operations in large volumes. But they’re limited to strikes with one-way attack drones like Shaheds and, occasionally evacuation of the wounded, not taking positions. The Lancet drones produced by Russia’s ZALA company are guided on final approach to their targets by matching camera imagery to preloaded maps. It works well enough—because Russian forces place less of a premium on collateral damage or striking the right target. For Ukrainians, the goal is greater autonomy, allowing one operator to preside over fleets of ground and air robots but with confidence that they will perform the mission assigned, hit the target that they’re supposed to hit and not simply whatever happens to be there when the drone finally arrives. It’s the same sort of complex multi-drone swarm capability that the Pentagon is seeking to develop. Ukrainian Air Force Capt. Max Maslii, deputy chief of staff for the 96th Anti-Aircraft Missile Brigade, described that goal as a departure from the way Russia operates “autonomous” drones like the Lancet, as isolated flying bombs. Under the “new paradigm,” Maslii told Defense One, the drones would be able to “find the … more efficient way to accomplish this mission, together with such machines.” At that point, he said, operators wouldn’t be stuck piloting one drone at a time. They would work more like technicians managing a larger, more complex system. “Our job will be … to produce a lot of drones, to put them in the proper place, to take care [of] the systems that manage those drones, and just to, you know, turn them on.” NEXT STORY: Put nuclear reactors in space within a few years, White House tells Pentagon Help us tailor content specifically for you: Thank you for subscribing! Please check out our other newsletter offerings on our Newsletter page.

All Headlines North Korea Sports Top News Most Viewed Korean Newspaper Headlines Today in Korean History Yonhap News Summary Editorials from Korean Dailies URL is copied. SEOUL, March 23 (Yonhap) -- HD Hyundai Co. said Monday its key affiliates and a U.S. robotics firm have forged a partnership to develop and commercialize artificial intelligence (AI)-powered humanoid welding robots for shipyard projects. The partnership agreement was signed recently between HD Korea Shipbuilding & Offshore Engineering (KSOE) Co. and HD Hyundai Robotics Co., along with Persona AI, a Houston-based company specializing in humanoid robots, according to the South Korean shipbuilder. It is a follow-up to a memorandum of understanding signed in May of last year on developing humanoid robots for shipyard welding. HD Hyundai noted that a prototype under development since last year has demonstrated sufficient technological feasibility and potential. Under the agreement, HD KSOE will develop welding training technologies for robots using data accumulated at shipyards, while HD Hyundai Robotics will oversee system integration for robot deployments. Persona AI plans to develop a bipedal humanoid platform capable of stable movement in shipyard environments. HD Hyundai said it plans to gradually deploy shipyard-specific humanoid welding robots at actual shipbuilding sites capable of performing complex tasks. This photo provided by HD Hyundai Co. on March 23, 2026, shows (from L to R) Song Young-hoon, head of the solutions development division at HD Hyundai Robotics Co., Lee Dong-joo, head of the manufacturing innovation institute at HD KSOE Co., and Persona AI Chief Executive Officer (CEO) Nick Radford posing for a commemorative photo at HD Hyundai Co.'s global research and development center in Pangyo, south of Seoul, after the companies signed a partnership agreement to develop AI-powered humanoid welding robots. (PHOTO NOT FOR SALE) (Yonhap) odissy@yna.co.kr(END) All News National North Korea Economy/Finance Biz Culture/K-pop Sports Images Videos Top News Most Viewed Korean Newspaper Headlines Today in Korean History Yonhap News Summary Editorials from Korean Dailies Korea in Brief Useful Links Weather Advertise with Yonhap News Agency

A McDonald’s in the Chinese megacity of Shanghai is testing humanoid robots in roles usually the preserve of human workers, with other types of robots also let loose inside the restaurant to greet and entertain diners. Truth be told, the robots don’t look particularly advanced, but a video (below) showing them in action does hint at a future where bipedal bots and other machines handle routine tasks at fast food restaurants, from welcoming customers and taking orders to delivering food and cleaning the floor. A McDonald’s in Shanghai has begun deploying humanoid robots (from KEENON Robotics) to serve customers.> These humanoid robots provide information, greet guests, and help enliven the atmosphere.> Food delivery robots serve meals to customers and collect used trays.in the… pic.twitter.com/IEFzucz3IE The McDonald’s trial, using robots supplied by Chinese firm Keenon Robotics, comes at a time of economic contradiction in China, where businesses in some sectors are struggling to hire even as millions of young people face difficulty finding work. It’s this tension that makes the McDonald’s trial stand out, with restaurant operators interested in deploying a reliable, potentially low-cost workforce in a strategy that raises fears of displacement among human workers in the service sector, which up to now has been a popular route into the workforce. The reality, however, is more complicated. China’s workforce is shrinking as the population ages, while many younger job seekers are reluctant to take on low-paid, repetitive work. In that case, robot technology could be used to fill gaps rather than simply replace people. Still, the presence of robots in such a visible, everyday setting highlights how quickly that balance could shift. While it could be a while before McDonald’s deploys humanoid robots in a more meaningful way, adding them to restaurants as greeters and entertainers could potentially draw curious diners, especially families with kids who might want to interact with the machines while waiting for their meal to arrive. Even if the fast food giant eventually wants robots to run its restaurants, such a scenario is almost certainly many years away, simply because the technology isn’t yet up to it. What feels more likely, at least in the short term, is a hybrid setup where human workers handle the majority of tasks while the robots take on more basic, customer-facing roles out front. Subtlety is overrated, and MSI just proved that. The Taiwanese laptop maker has rolled out a sweeping refresh, unveiling more than a dozen new gaming laptops spread across its Cyborg, Crosshair, Raider, Stealth, and Titan lineups. The models cover 15-inch, 16-inch, and 18-inch form factors, ensuring there’s something for every gamer or professional user, making it hard for buyers to run out of excuses for not upgrading this year. Google made an unexpected cameo on Macs with the launch of a native Gemini app. What’s even more interesting (and a bit funny) is that the app arrived at Apple’s long-promised Siri upgrade (and a rumored standalone app for the voice assistant). The free app is available on macOS 15 and above. Though the app isn’t available on the App Store (yet), you can download it from Google’s official landing page. Nothing launched a genuinely useful app called Warp earlier today, with a simple idea: allowing Android users to share files, links, and copied texts directly to their Mac, Windows, or Linux machines without including any cables or convoluted workarounds. Nothing announced the app for Chrome and Edge (Chromium-based web browsers) and Android smartphones, floating it on both the Chrome Web Store and the Google Play Store (via 9To5Google). However, a few hours later, the app is nowhere to be found, with the official listings returning errors. Upgrade your lifestyleDigital Trends helps readers keep tabs on the fast-paced world of tech with all the latest news, fun product reviews, insightful editorials, and one-of-a-kind sneak peeks.

SEOUL, March 23 (UPI) -- South Korea's HD Hyundai said Monday it would test welding humanoid robots at shipyards operated by its affiliates, including the world's leading shipbuilder HD Hyundai Heavy Industries. HD Hyundai noted that its subsidiaries have recently signed an agreement with U.S.-based artificial intelligence company Persona AI, a high-profile startup on industrial humanoid robots. Under the partnership, HD Hyundai will leverage its shipyard data to come up with robot training technologies for field testing. Meanwhile, Persona AI is poised to focus on developing a bipedal humanoid platform designed to stably move in complex shipyard environments, according to HD Hyundai. The Seoul-based conglomerate said that the project aims to test robots capable of performing high-level tasks such as welding by replicating the expertise and working patterns of highly skilled personnel. "Humanoids tailored for shipyards will serve as a key foundation for future smart facilities by enhancing worker safety while improving production efficiency," HD Hyundai said in a statement. "We plan to lead a new paradigm in the shipbuilding industry by introducing humanoids into ship construction sites." The group did not disclose a timeline for deploying the robots in actual operations. Such a move is expected to face strong opposition from labor unions. The share price of HD Hyundai dipped 9.23% on Monday on the Seoul bourse. The country's benchmark KOSPI dropped 6.49% amid rising tensions between Washington and Tehran. Read More Labor union rallies behind Korea Zinc before key shareholder battle Korea Aerospace Industries' new CEO takes office South Korea seeks to attract global visitors with 'K-Chicken Belt' Topics BusinessTechnology Latest Headlines World News // 21 minutes ago South Korea pet insurance market grows but uptake remains low April 15 (Asia Today) -- S. Korea's pet insurance market has expanded more than threefold in the past three years, but low enrollment rates continue to limit its growth. World News // 26 minutes ago IAEA chief says North Korea expands uranium enrichment April 15 (Asia Today) -- Rafael Grossi said that North Korea has built a new uranium enrichment facility, signaling a significant expansion of its nuclear capabilities. World News // 33 minutes ago South Korea import prices post biggest jump in 28 years April 15 (Asia Today) -- S. Korea's import prices rose 16.1% in March from a month earlier, the sharpest monthly increase in more than 28 years, according to the Bank of Korea. World News // 45 minutes ago South Korea moves to stabilize farm supplies amid price risks April 15 (Asia Today) -- S. Korea has secured stable supplies of agricultural fertilizer through July and is expanding subsidies to offset rising costs farming materials. World News // 54 minutes ago South Korean charities push tax incentives for legacy giving April 15 (Asia Today) -- More than 200 charities in South Korea are urging lawmakers to adopt tax incentives for legacy donations to expand the country's culture of giving. World News // 1 hour ago North Korea avoids 'Day of the Sun' term in state media April 15 (Asia Today) -- N. Korea has avoided using the term "Day of the Sun" in state media coverage marking the birthday of Kim Il Sung, signaling a shift to current leader. World News // 1 hour ago South Korea pushes looser rules for high-tech sectors April 15 (Asia Today) -- Lee Jae-myung said South Korea should shift to a "negative regulation" system in advanced technology sectors to strengthen global competitiveness. World News // 12 hours ago Iran threatens shipping in Red Sea as Trump says talks likely to restart April 15 (UPI) -- As a U.S. blockade of Iranian ports continues, Iran threatened Wednesday to halt shipping in the Red Sea, the Persian Gulf and the Gulf of Oman. World News // 5 hours ago Milei's approval falls in Argentina as inflation picks up again BUENOS AIRES, April 15 (UPI) -- Public discontent with Argentine President Javier Milei is rising as inflation accelerates again and many people say they have yet to feel the benefits of the government's economic reforms. World News // 5 hours ago Separatists in Cameroon pause fighting for Pope Leo XIV visit April 15 (UPI) -- Pope Leo XIV landed in Cameroon Wednesday, and English-speaking separatists in the country announced a pause in fighting for "safe travel passage." SEOUL, March 23 (UPI) -- South Korea's HD Hyundai said Monday it would test welding humanoid robots at shipyards operated by its affiliates, including the world's leading shipbuilder HD Hyundai Heavy Industries. HD Hyundai noted that its subsidiaries have recently signed an agreement with U.S.-based artificial intelligence company Persona AI, a high-profile startup on industrial humanoid robots. Under the partnership, HD Hyundai will leverage its shipyard data to come up with robot training technologies for field testing. Meanwhile, Persona AI is poised to focus on developing a bipedal humanoid platform designed to stably move in complex shipyard environments, according to HD Hyundai. The Seoul-based conglomerate said that the project aims to test robots capable of performing high-level tasks such as welding by replicating the expertise and working patterns of highly skilled personnel. "Humanoids tailored for shipyards will serve as a key foundation for future smart facilities by enhancing worker safety while improving production efficiency," HD Hyundai said in a statement. "We plan to lead a new paradigm in the shipbuilding industry by introducing humanoids into ship construction sites." The group did not disclose a timeline for deploying the robots in actual operations. Such a move is expected to face strong opposition from labor unions. The share price of HD Hyundai dipped 9.23% on Monday on the Seoul bourse. The country's benchmark KOSPI dropped 6.49% amid rising tensions between Washington and Tehran.

SEOUL, March 23 (UPI) -- South Korea's HD Hyundai said Monday it would test welding humanoid robots at shipyards operated by its affiliates, including the world's leading shipbuilder HD Hyundai Heavy Industries. HD Hyundai noted that its subsidiaries have recently signed an agreement with U.S.-based artificial intelligence company Persona AI, a high-profile startup on industrial humanoid robots. Under the partnership, HD Hyundai will leverage its shipyard data to come up with robot training technologies for field testing. Meanwhile, Persona AI is poised to focus on developing a bipedal humanoid platform designed to stably move in complex shipyard environments, according to HD Hyundai. The Seoul-based conglomerate said that the project aims to test robots capable of performing high-level tasks such as welding by replicating the expertise and working patterns of highly skilled personnel. "Humanoids tailored for shipyards will serve as a key foundation for future smart facilities by enhancing worker safety while improving production efficiency," HD Hyundai said in a statement. "We plan to lead a new paradigm in the shipbuilding industry by introducing humanoids into ship construction sites." The group did not disclose a timeline for deploying the robots in actual operations. Such a move is expected to face strong opposition from labor unions. The share price of HD Hyundai dipped 9.23% on Monday on the Seoul bourse. The country's benchmark KOSPI dropped 6.49% amid rising tensions between Washington and Tehran. Read More Labor union rallies behind Korea Zinc before key shareholder battle Korea Aerospace Industries' new CEO takes office South Korea seeks to attract global visitors with 'K-Chicken Belt' Topics BusinessTechnology Latest Headlines World News // 21 minutes ago South Korea pet insurance market grows but uptake remains low April 15 (Asia Today) -- S. Korea's pet insurance market has expanded more than threefold in the past three years, but low enrollment rates continue to limit its growth. World News // 26 minutes ago IAEA chief says North Korea expands uranium enrichment April 15 (Asia Today) -- Rafael Grossi said that North Korea has built a new uranium enrichment facility, signaling a significant expansion of its nuclear capabilities. World News // 33 minutes ago South Korea import prices post biggest jump in 28 years April 15 (Asia Today) -- S. Korea's import prices rose 16.1% in March from a month earlier, the sharpest monthly increase in more than 28 years, according to the Bank of Korea. World News // 45 minutes ago South Korea moves to stabilize farm supplies amid price risks April 15 (Asia Today) -- S. Korea has secured stable supplies of agricultural fertilizer through July and is expanding subsidies to offset rising costs farming materials. World News // 54 minutes ago South Korean charities push tax incentives for legacy giving April 15 (Asia Today) -- More than 200 charities in South Korea are urging lawmakers to adopt tax incentives for legacy donations to expand the country's culture of giving. World News // 1 hour ago North Korea avoids 'Day of the Sun' term in state media April 15 (Asia Today) -- N. Korea has avoided using the term "Day of the Sun" in state media coverage marking the birthday of Kim Il Sung, signaling a shift to current leader. World News // 1 hour ago South Korea pushes looser rules for high-tech sectors April 15 (Asia Today) -- Lee Jae-myung said South Korea should shift to a "negative regulation" system in advanced technology sectors to strengthen global competitiveness. World News // 12 hours ago Iran threatens shipping in Red Sea as Trump says talks likely to restart April 15 (UPI) -- As a U.S. blockade of Iranian ports continues, Iran threatened Wednesday to halt shipping in the Red Sea, the Persian Gulf and the Gulf of Oman. World News // 5 hours ago Milei's approval falls in Argentina as inflation picks up again BUENOS AIRES, April 15 (UPI) -- Public discontent with Argentine President Javier Milei is rising as inflation accelerates again and many people say they have yet to feel the benefits of the government's economic reforms. World News // 5 hours ago Separatists in Cameroon pause fighting for Pope Leo XIV visit April 15 (UPI) -- Pope Leo XIV landed in Cameroon Wednesday, and English-speaking separatists in the country announced a pause in fighting for "safe travel passage." SEOUL, March 23 (UPI) -- South Korea's HD Hyundai said Monday it would test welding humanoid robots at shipyards operated by its affiliates, including the world's leading shipbuilder HD Hyundai Heavy Industries. HD Hyundai noted that its subsidiaries have recently signed an agreement with U.S.-based artificial intelligence company Persona AI, a high-profile startup on industrial humanoid robots. Under the partnership, HD Hyundai will leverage its shipyard data to come up with robot training technologies for field testing. Meanwhile, Persona AI is poised to focus on developing a bipedal humanoid platform designed to stably move in complex shipyard environments, according to HD Hyundai. The Seoul-based conglomerate said that the project aims to test robots capable of performing high-level tasks such as welding by replicating the expertise and working patterns of highly skilled personnel. "Humanoids tailored for shipyards will serve as a key foundation for future smart facilities by enhancing worker safety while improving production efficiency," HD Hyundai said in a statement. "We plan to lead a new paradigm in the shipbuilding industry by introducing humanoids into ship construction sites." The group did not disclose a timeline for deploying the robots in actual operations. Such a move is expected to face strong opposition from labor unions. The share price of HD Hyundai dipped 9.23% on Monday on the Seoul bourse. The country's benchmark KOSPI dropped 6.49% amid rising tensions between Washington and Tehran.

An editorially independent publication supported by the Simons Foundation. Get the latest news delivered to your inbox. Create a reading list by clicking the Read Later icon next to the articles you wish to save. Type search term(s) and press enter Popular Searches March 13, 2026 Companies are developing and now promoting a future full of humanoid robots. Henry Flores for Quanta Magazine Contributing Writer March 13, 2026 The last time I covered the science of humanoid robots, the state of the art looked downright Orwellian — by which I mean, “four legs good, two legs bad.” It was 2015. Boston Dynamics’ first “Spot” quadruped had taken YouTube by storm, confidently trotting up stairs and recovering from vicious kicks. Also popular at the time: humanoids falling down. Constantly. I felt sorrier for those tottering metal lobsters than I ever did for Spot. Bipedal locomotion is hard. Cut to now. Humanoids have apparently become so advanced that Tesla is mothballing some electric car models to make way for its Optimus humanoid robot, and start-ups are preselling android butlers with a straight face. Hype aside, I was genuinely curious: Did a paradigm shift happen in the field when I wasn’t looking? Sure, “AI” happened (that is, in the post-ChatGPT sense). I certainly hadn’t overlooked that. But I had no idea what it possibly had to do with robots not falling down anymore. AI breakthroughs have made humanoid robots more capable than ever. But they still struggle with everyday tasks like stairs and doors. For a reality check, I called Scott Kuindersma, who recently left Boston Dynamics after many years there, and Jonathan Hurst of Agility Robotics. Both scientists had been present and involved during the robot-faceplant days. Surely today’s robotic bipedal marvels can ascend a few stairs and open a door without breaking a nonexistent sweat, something they famously struggled with a decade ago. I asked each researcher: Can your flagship robot — Boston Dynamics’ Atlas or Agility’s Digit, two of the most credible and pedigreed humanoids on Earth — handle any set of stairs or doorway? “Not reliably,” Hurst said. “I don’t think it’s totally solved,” Kuindersma said. Don’t get me wrong: I don’t believe that some sock-faced robot zombie is close to taking over my household chores. But stairs and doors? It’s 2026. Why are humanoids still this … hard? To be fair, a paradigm shift did happen. Three, actually. First, deep learning — neural networks running on fast GPU chips — turbocharged computer vision and reinforcement learning, which radically improved the speed and sophistication with which robots could perceive and interact with their environments. Then in 2016, a revolution in actuation (roboticist-speak for “making parts move”) began: Heavy hydraulic mechanisms were replaced by smaller, “proprioceptive” electric motors that gave legged robots animal-like nimbleness. Most recently came the large language models. Adapting chatbot technology for robots, it turns out, lets them autonomously plan and perform multistep tasks, such as coring an apple or emptying a dishwasher (in demos, at least). In philosophy, “qualia” refers to the subjective qualities of our experience: what it’s like for Alice to see blue or for Bob to feel delighted. Qualia are “the ways things seem to us,” as the late philosopher Daniel Dennett put it. In these essays, our columnists follow their curiosity, and explore important but not necessarily answerable scientific questions. These advances created the night-and-day difference between “Running Man,” the hulking, halting version of Atlas that won second place in 2015’s DARPA Robotics Challenge, and the svelte, smooth Atlas recently shown breakdancing and autonomously moving irregular items from one bin to another (while dealing with interference from a hockey stick–wielding human). That fluid gait, for example, comes from deep reinforcement learning. Roboticists once coordinated each movement with various hand-engineered algorithms, using equations to model the (simplified) physics of the robot. Now they train neural networks to act as “whole-body controllers” by running countless digital simulations of the humanoid. This process teaches the network a “policy” for how to translate feedback from its environment into actions. “We use reinforcement learning to build a policy that’s handling the body coordination, collision avoidance, balance, all that stuff,” Kuindersma said. There’s no longer any need to model a robot’s leg as a linear inverted pendulum, for example. “That’s just gone by the wayside,” he said. Get Quanta Magazine delivered to your inbox The Atlas robot from Boston Dynamics shows off in a video from early 2025. Boston Dynamics/Anadolu Agency via Getty Images This strategy was aided by the proprioceptive actuators pioneered by Sangbae Kim of the Massachusetts Institute of Technology in his Cheetah series of robots. “Reinforcement learning has existed for a long time, you know. People tried it before,” Kim said. “But if you use conventional [motors], the robot just breaks” every time it fails to perfectly execute a policy in the real world — or encounters an obstacle or disturbance. Kim’s actuators got around the problem with controllable “compliance,” or flexible springiness. Over the past decade, they’ve gotten cheaper and more widely accessible. “Reinforcement learning solved a lot of the [bipedal] locomotion problem, but the hardware was the enabler,” Kim said. To have robots which work like humans, I think we have to master physics. Pulkit Agrawal If reinforcement learning and compliant actuation were gifts to humanoid robotics, multimodal AI put a bow on it. In 2023, Google DeepMind introduced “vision-language-action” (VLA) models, which can take in video and natural language and produce movement commands as outputs. “If you say ‘I’m thirsty,’ it knows you probably want to drink, and it can [generate] the steps that [the robot] needs to take: Go find a thing, and then pick it up in this way,” said Carolina Parada, head of robotics at Google DeepMind. “This is something that, before three years ago, you would have to go hard-code.” In a stroke, VLAs united previously disparate approaches to robotic perception, planning, and control into one general-purpose pipeline. Robust embodiment, check. Generalizable intelligence, check. (A start, anyway.) So why don’t they add up to humanoids being scientifically “solved” — at least in principle? Pulkit Agrawal, who studies robot learning at the appropriately named Improbable AI Lab at MIT, had an answer when I reached him there last month. “To have robots which work like humans,” he said, “I think we have to master physics.” He wasn’t referring to cosmic matters like general relativity or quantum gravity, nor to the virtual “world models” that currently excite leading AI researchers such as Yann LeCun. Instead, Agrawal is talking about mastering something a high school science student ought to be familiar with: force and inertia. Press images of the Neo from 1X (left) and Tesla’s Optimus (right) imagine a future of humanoid helpers. Courtesy of 1X; Tesla The whole point of the humanoid form factor, after all, is to deliver what Kim calls “multipurpose mobile manipulation,” or the ability to move almost anywhere (including on stairs and through doors) and handle almost anything (from unloading pallets to screwing in light bulbs), without hurting anyone in the process. In short, what we do every day. “These things are about [controlling] forces, if you want to do them at speeds of a human,” Agrawal said. “Force control has been a thing in classical [robotics]. But in modern machine learning land, it’s not been that widespread.” Force control is simple in principle. Picture a robot arm drawing on a whiteboard — without smashing the tip of the marker. Roboticists have known how to make this happen for more than 40 years: They program the arm to behave as if it has an imaginary spring and shock absorber attached to it. “One can make the spring really soft in the direction pointing into the whiteboard, and stiffer along the surface of the whiteboard,” Kuindersma said. “That way the robot maintains the right pressure with the marker while precisely writing the lines and curves of the letters.” This feedback can be driven by force sensors built into the robot’s joints, but the catch is that the classical approaches require a lot of knowledge about the robot, environment, and task in order to work, he further explained. That approach to controlling force works great for industrial robots with specific tasks to perform, and it even helped with humanoid locomotion. But it was impossible to generalize. Kim’s proprioceptive electric actuators, also called quasi-direct drive actuators, simplified things. Not only were they designed to absorb unexpected impacts without damage, they were also very “transparent,” which meant that the motor converted electrical current into a proportional amount of force (and vice versa) with relatively little error. In essence, the motor itself became a force sensor, which meant “you can remove cost and complexity from your robot by eliminating dedicated force sensors,” Kuindersma said. As reinforcement learning eclipsed manual programming as a way of controlling humanoid movement, “classic” force control was not forgotten. It just got abstracted and delegated, in a way, to both hardware and AI. “From an AI point of view, it’s not like you have to be thinking about force control,” Hurst said. “It’s more like you kind of know that you need a quasi-direct drive motor to get close [to the force regulation necessary], then put [the neural network] in simulation and iterate a million times — and then you can put it on the robot and get cool behaviors.” Those neural networks are learning generalized policies that control the positions of a robot’s body parts. Force regulation often happens only indirectly in simulation training, or sometimes as a side effect when learned from video or human input. But those methods don’t explicitly teach the physics of force — at least, not yet. “A lot of the signals that are required for doing intelligent force control are not present in [video and human demonstration] data,” Kuindersma said. DeepMind’s Parada acknowledged that the VLA models basically just learn to move between specifically defined poses — and this approach goes a long way. “We’ve been surprised ourselves at how far you can push it, without any other sensing,” she said. In 2015, the most advanced humanoid robots in the world competed at the DARPA Robotics Challenge Finals. The tech has since improved. DARPA But only so far. As long as robot bodies remain relatively stiff and heavy compared to ours, “they have high inertia, and they’re not [as] compliant,” Agrawal said, which means that without force control, they will struggle with precision tasks in complicated environments. “If you’re going to touch delicate objects and you have small errors, bad things are going to happen.” Picture a regular egg and another made of solid steel: One of them needs to be picked up much more carefully. One way to get around this problem, used by many impressive systems alongside positional accuracy, is just to go slow. Imagine trying to move a chair with your car, Agrawal said: “If I go slowly, I can be precise on how I move [my position], and then I can control where the chair goes, so the [force] problem goes away.” That’s part of why Atlas moves like molasses while grasping auto parts but glides like a gymnast when it’s not touching anything except the floor. “It would be an overstatement to say that force control is absolutely required in every useful manipulation task — that’s just not true,” Kuindersma said. But he, Hurst, and Parada all readily grant that clever force workarounds won’t deliver the all-purpose mobile dexterity our robot butlers need. Even if today’s VLA-brained bots, refined by reinforcement learning, had “an internet-sized” amount of positional data to train on, “it’s very likely you [would] have to do some additional work,” Parada said. “Humans feel the forces that are working against you when you’re trying to open a bottle.” Humanoids, for the most part, still don’t, which means they have not mastered physics — at least not in the way we have, from a lifetime of interacting with our environments through the extraordinarily complex musculoskeletal and nervous systems gifted to us by evolution. That’s a big reason why even doors and stairs aren’t fully “solved” for present-day humanoids. These stairs, that door? Probably. But all stairs and doors, plus everything else? “There’s no world in which there are actually useful, autonomous [humanoid] robots that are only doing position-based control,” Kuindersma said. “Force as a first-class citizen is absolutely required.” So how do we get over the wall, scientifically speaking? Most of the experts I asked suspect that it will take a new blend of hardware and software advances. Tactile sensors for better data collection and robot hands that combine high power, compliance, and transparency with low inertia would accomplish a lot, and nobody believes that true material breakthroughs (like replacing motors with artificial muscles) will be necessary. “The hardware is exceptional, and if you’re blaming [it], you’re making excuses,” said Russ Tedrake, another longtime MIT roboticist I spoke to. “If you put a human brain through the hardware we have today — by teleoperating it, for instance — it’s incredibly capable.” Finding more intelligent ways to control it is key. The Digit robot from Agility Robotics demonstrates fine motor control in an unstructured environment. Agility Robotics When asked how to achieve that, everyone had a different answer. Agrawal is studying how to combine force control with reinforcement learning by having humanoids learn compliant behaviors in simulation, instead of moving between rigidly defined positions. Tedrake, whose work on “large behavior models” (a cousin of VLAs) produced the apple-coring robot demo, recently argued in Science Robotics for a ChatGPT-style regime of “large-scale data collection and large pretrained models.” Frank Park, who wrote the book on modern robotics — literally, the textbook titled Modern Robotics — believes that current AI approaches should be torn down to the studs and replaced with ones that make physics fundamentals (such as force and acceleration) learnable at a foundational level. “The VLA architecture is just all wrong,” he told me. “I believe that approach is doomed to fail.” In all these conversations, what struck me most wasn’t the debates about which kinds of sensors, data, or AI architecture could “solve” humanoid robotics. Rather, it was the sense that the scientific ethos of the field had changed. Hurst, who had just spun Agility Robotics out of his Oregon State University lab when we first spoke, put a fine point on it. “I remember Gill Pratt, who was the director of the MIT Leg Lab and then the program manager for the DARPA Robotics Challenge, saying that his big worry was that we’d end up using reinforcement learning and AI to make robots walk and run before we ever actually understood how it works,” he said. “And in a lot of ways, we’re kind of doing that.” (Editor’s note: Gill Pratt recalled this conversation differently. He acknowledged that machine learning could allow performance beyond our formal understanding, but not that this was a cause for worry.) Tedrake agreed but said that it’s hardly the first time we’ve taken scientific and engineering leaps without a firm grip on the fundamentals. “If you look at electricity and magnetism, there was the Volta stage where you’re sticking electrodes in frogs,” he said. “And then we had Faraday, who did exactly the right experiments, and then eventually we had Maxwell tell us the governing equations. I think we’re in the Volta stage.” So when will humanoids be solved? “Robots are still bad, and it will take time. But the bones are good. Both are true,” Tedrake said. “And it’s still hard.” Contributing Writer March 13, 2026 Get Quanta Magazine delivered to your inbox Get highlights of the most important news delivered to your email inbox Quanta Magazine moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (New York time) and can only accept comments written in English. Forgot your password ? We’ll email you instructions to reset your password Enter your new password

From daily news and career tips to monthly insights on AI, sustainability, software, and more—pick what matters and get it in your inbox. Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. Follow Us On Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. The partnership targets robots for welding, mobility and precision tasks, with phased rollout across shipyard operations. Hyundai has partnered with US-based robotics firm Persona AI to develop and commercialize humanoid welding humanoids for shipyards. A joint development agreement was signed by HD Hyundai with HD Korea Shipbuilding & Offshore Engineering, HD Hyundai Robotics, and Persona AI. Under the agreement, HD KSOE will develop welding training systems using shipyard data, while HD Hyundai Robotics will handle integration. Persona AI will design a bipedal humanoid platform, with phased deployment planned across shipbuilding sites. The deal builds on a May 2025 partnership after successful prototype evaluations, which aimed to develop humanoid robots capable of performing advanced welding tasks in shipyards. Growing labor shortages in heavy industry, particularly in high-risk tasks such as welding, are increasing the urgency for rugged, autonomous humanoid robots. Aiding such a transition, HD Hyundai announced a joint development agreement on March 23, and the signing ceremony took place at the HD Hyundai Global R&D Center in South Korea, marking a step forward in efforts to automate complex shipbuilding processes. An earlier agreement was set in May 2025, following which successful evaluations of a humanoid prototype’s technical feasibility and real-world applicability were conducted. Under the deal, HD Korea Shipbuilding & Offshore Engineering will develop artificial intelligence-based welding training systems using data collected from shipyard operations and integrate them into production workflows. HD Hyundai Robotics will oversee system integration, including quality analysis, control technologies, and field testing. Persona AI will focus on developing a bipedal humanoid platform capable of stable movement in challenging shipyard environments, reports The Korea Times. The collaboration aims to produce robots capable of performing high-skill tasks such as welding, mobility, perception, and precision control, with gradual deployment planned across shipyard operations. A prototype is targeted for completion by late 2026, followed by field testing and commercial deployment in 2027. The collaboration represents a significant step toward building smart shipyards where humans and robots operate side by side. Robotics player Persona sees the partnership with HD Hyundai and its affiliates as a significant step beyond a symbolic collaboration, noting that shipyards are among the largest real-world testing environments for deploying and validating durable humanoid robotic systems. Persona is positioning humanoid robots as a solution to skilled labor shortages in demanding industrial sectors. Its systems are designed for high-intensity environments and focus on “3D” tasks—dull, dirty, and dangerous—commonly found in shipyards, construction, and energy infrastructure, reducing the physical strain on human workers. The company highlights a technological foundation influenced by advanced robotics developed through NASA, combining this legacy with practical engineering aimed at real-world deployment. Central to its approach is a modular humanoid platform equipped with a highly dexterous robotic hand derived from NASA-linked intellectual property, enabling precise work in complex, unstructured settings. The platform uses interchangeable “Personas” that allow it to adapt across industries and tasks. In shipbuilding, the robots are designed for confined-space operations, hull welding and repair work, where workforce attrition in key trades can exceed 30 percent. In the energy sector, they support pipe welding, inspection, and maintenance as aging labor pools and automation reshape operations. The company aims to deliver scalable, reliable labor through continuous operation, improved efficiency, and reduced rework, advancing automation in heavy industry. Jijo is an automotive and business journalist based in India. Armed with a BA in History (Honors) from St. Stephen's College, Delhi University, and a PG diploma in Journalism from the Indian Institute of Mass Communication, Delhi, he has worked for news agencies, national newspapers, and automotive magazines. In his spare time, he likes to go off-roading, engage in political discourse, travel, and teach languages. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Premium Follow

Компания Google заключила партнёрское соглашение с немецкой компанией Agile Robots — поисковый гигант решил сделать ставку на робототехнику как ключевой инструмент развития по направлению искусственного интеллекта. Источник изображения: agile-robots.com Agile Robots специализируется на разработке интеллектуальных роботизированных манипуляторов и оснащённых сенсорами человекоподобных роботов. В рамках сотрудничества в оборудование немецкой компании будут интегрироваться ИИ-модели Google Gemini Robotics. «Партнёрство основывается на убеждении, что применение ИИ в физическом мире должно трансформироваться. Объединив оборудование Agile Robots и другие разрабатываемые в Германии решения в области робототехники с базовыми моделями Google DeepMind Gemini Robotics, обе стороны добьются успеха за счёт развёртывания роботов, сбора данных, обучения моделей и поэтапного совершенствования», — говорится в блоге компании. Google будет получать данные о работе своих продуктов в реальном мире — для технологического гиганта робототехника выступает одним из важнейших сценариев применения ИИ, где она конкурирует с Amazon и Tesla. У компании множество партнёрских соглашений по этому направлению. Мюнхенская Agile Robots к настоящему моменту развернула более 20 000 роботизированных систем по всему миру; решения Google она намеревается масштабно интегрировать в уже действующие системы. На начальном этапе это будут «высокоценные промышленные» сценарии, в том числе в производстве. Компания поможет Google и далее разрабатывать «более совершенные модели ИИ для роботов нового поколения». В прошлом году Google выпустила базовую и рассуждающую ИИ-модели Gemini Robotics и объявила о сотрудничестве с техасской Apptronik; в этом году стало известно, что подразделение Google DeepMind будет сотрудничать с Boston Dynamics в работе над роботом Atlas. Под управление Google была переведена входившая в холдинг Alphabet компания Intrinsic, которой прочат судьбу «Android в робототехнике»; в DeepMind также приняли на работу бывшего технического директора Boston Dynamics Аарона Сондерса (Aaron Saunders). Впрочем, некоторые сотрудники Google не вполне довольны сотрудничеством поискового гиганта с Boston Dynamics — у компании есть действующие контракты с Министерством обороны США. Источник: Укажите имя пользователя: и пароль: Войти © 1997—2026 Электронное периодическое издание "3ДНьюс" | Свидетельство о регистрации СМИ Эл ФС 77-22224 выдано Федеральной Службой по надзору за соблюдением законодательства в сфере массовых коммуникаций и охране культурного наследия При цитировании документа ссылка на сайт с указанием автора обязательна. Полное заимствование документа является нарушениемроссийского и международного законодательства и возможно только с согласия редакции 3DNews. Во время посещения сайта вы соглашаетесь с использованием нами файлов cookie, метрических программ, Пользовательским соглашением и даёте согласие на обработку и трансграничную передачу персональных данных.

From daily news and career tips to monthly insights on AI, sustainability, software, and more—pick what matters and get it in your inbox. Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. Follow Us On Access expert insights, exclusive content, and a deeper dive into engineering and innovation all with fewer ads or a completely ad-free experience. All Rights Reserved, IE Media, Inc. OpenMind’s new robot app store allows humanoid and quadruped robots to expand their skills through apps. OpenMind, a robotics software company, has launched a new robot app store designed to let humanoid and quadruped robots gain new skills and abilities. The platform is live now and aims to eventually host thousands of apps that can expand the capabilities of robots beyond their built-in hardware. According to a Forbes report, the app store is built on OM1, OpenMind’s modular operating system, which allows developers to create apps that package specific skills for distribution across multiple robot platforms. OpenMind is working with partners including UBbtech, Agibot, Deep Robotics, Fourier, Booster, Dobot, LimX, and Magic Lab. “Computers and phones come with an operating system to provide the basics, but the real magic is the ability for everyone to personalize their phones and computers through apps and programs,” said Jan Liphardt, founder and CEO of OpenMind. “That’s how generic hardware comes to life and becomes your phone and your laptop. Your humanoid will be no different: thousands of apps, each representing skills from nursing and math education to cleaning and home safety, will give you almost unlimited choices.” Current apps cover a range of practical and experimental functions, from companionship, elder care, and home security to novelty apps like selfie-taking robots. The company expects the quality and complexity of apps to grow over time, similar to the early days of smartphone app stores. The platform emphasizes that software can evolve independently of hardware, enabling robots to learn new tasks and improve over time. Liphardt said, “Robots need a skill and cognition layer that evolves faster than hardware. The App Store is how robots become universal platforms whose skills can change over time to fit your needs.” OpenMind’s initial app catalog includes Omni-Guardian, which turns a robot into a companion and sentry capable of detecting intruders; Nova, which listens, sees, and moves to assist with daily tasks; WALL-E, which monitors digital assets and social feeds; Luckandroll OM1, enabling robots to interact with humans and coordinate with other robots; and Guardian, which can follow a user and take selfies. While some apps are experimental or low-effort, the approach mirrors the early days of iOS and Android app stores, where quirky and test apps dominated before the ecosystem matured. Liphardt notes that practical apps, such as floor cleaning or laundry assistance, will appear as robotics capabilities advance. The OpenMind developer ecosystem already includes over 1,000 developers worldwide and is open to additional developers and robot manufacturers. The company expects the store to expand rapidly as more apps and partners join, providing a growing library of skills for commercial and personal robots. By creating a software-focused platform, OpenMind is pushing the robotics industry toward modularity, flexibility, and user-driven innovation. This launch represents a significant step toward turning robots into universal, upgradable machines capable of performing diverse tasks in homes and workplaces. With over a decade-long career in journalism, Neetika Walter has worked with The Economic Times, ANI, and Hindustan Times, covering politics, business, technology, and the clean energy sector. Passionate about contemporary culture, books, poetry, and storytelling, she brings depth and insight to her writing. When she isn’t chasing stories, she’s likely lost in a book or enjoying the company of her dogs. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Premium Follow

Компания Google заключила партнёрское соглашение с немецкой компанией Agile Robots — поисковый гигант решил сделать ставку на робототехнику как ключевой инструмент развития по направлению искусственного интеллекта. Источник изображения: agile-robots.com Agile Robots специализируется на разработке интеллектуальных роботизированных манипуляторов и оснащённых сенсорами человекоподобных роботов. В рамках сотрудничества в оборудование немецкой компании будут интегрироваться ИИ-модели Google Gemini Robotics. «Партнёрство основывается на убеждении, что применение ИИ в физическом мире должно трансформироваться. Объединив оборудование Agile Robots и другие разрабатываемые в Германии решения в области робототехники с базовыми моделями Google DeepMind Gemini Robotics, обе стороны добьются успеха за счёт развёртывания роботов, сбора данных, обучения моделей и поэтапного совершенствования», — говорится в блоге компании. Google будет получать данные о работе своих продуктов в реальном мире — для технологического гиганта робототехника выступает одним из важнейших сценариев применения ИИ, где она конкурирует с Amazon и Tesla. У компании множество партнёрских соглашений по этому направлению. Мюнхенская Agile Robots к настоящему моменту развернула более 20 000 роботизированных систем по всему миру; решения Google она намеревается масштабно интегрировать в уже действующие системы. На начальном этапе это будут «высокоценные промышленные» сценарии, в том числе в производстве. Компания поможет Google и далее разрабатывать «более совершенные модели ИИ для роботов нового поколения». В прошлом году Google выпустила базовую и рассуждающую ИИ-модели Gemini Robotics и объявила о сотрудничестве с техасской Apptronik; в этом году стало известно, что подразделение Google DeepMind будет сотрудничать с Boston Dynamics в работе над роботом Atlas. Под управление Google была переведена входившая в холдинг Alphabet компания Intrinsic, которой прочат судьбу «Android в робототехнике»; в DeepMind также приняли на работу бывшего технического директора Boston Dynamics Аарона Сондерса (Aaron Saunders). Впрочем, некоторые сотрудники Google не вполне довольны сотрудничеством поискового гиганта с Boston Dynamics — у компании есть действующие контракты с Министерством обороны США. © 3DNews

Claire chatted to Kevin McAleer from kevsrobots about how to get started building robots at home. Kevin McAleer is a hobbyist robotics fanatic who likes to build robots, share videos about them on YouTube and teach people how to do the same. Kev has been building robots since 2019, when he got his first 3d printer and wanted to make more interesting builds. Kev has a degree in Computer Science, and because his day job is relatively hands-off, this hobby allows his creativity to have an outlet. Kev is a huge fan of Python and Micropython for embedded devices, and has a website – kevsrobots.com where you can learn more about how to get started in robotics.

People in Nigeria and India are strapping iPhones onto their heads and recording themselves doing chores. When Zeus, a medical student living in a hilltop city in central Nigeria, returns to his studio apartment from a long day at the hospital, he turns on his ring light, straps his iPhone to his forehead, and starts recording himself. He raises his hands in front of him like a sleepwalker and puts a sheet on his bed. He moves slowly and carefully to make sure his hands stay within the camera frame. Zeus is a data recorder for Micro1, a US company based in Palo Alto, California that collects real-world data to sell to robotics companies. As companies like Tesla, Figure AI, and Agility Robotics race to build humanoids—robots designed to resemble and move like humans in factories and homes—videos recorded by gig workers like Zeus are becoming the hottest new way to train them. Micro1 has hired thousands of contract workers in more than 50 countries, including India, Nigeria, and Argentina, where swathes of tech-savvy young people are looking for jobs. They’re mounting iPhones on their heads and recording themselves folding laundry, washing dishes, and cooking. The job pays well by local standards and is boosting local economies, but it raises thorny questions around privacy and informed consent. And the work can be challenging at times—and weird. Zeus found the job in November, when people started talking about it everywhere on LinkedIn and YouTube. “This would be a real nice opportunity to set a mark and give data that will be used to train robots in the future,” he thought. Zeus is paid $15 an hour, which is good income in Nigeria’s strained economy with high unemployment rates. But as a bright-eyed student dreaming of becoming a doctor, he finds ironing his clothes for hours every day boring. “I really [do] not like it so much,” he says. “I’m the kind of person that requires … a technical job that requires me to think.” Zeus, and all the workers interviewed by MIT Technology Review, asked to be referred to only by pseudonyms because they were not authorized to talk about their work. Humanoid robots are notoriously hard to build because manipulating physical objects is a difficult skill to master. But the rise of large language models underlying chatbots like ChatGPT has inspired a paradigm shift in robotics. Just as large language models learned to generate words by being trained on vast troves of text scraped from the internet, many researchers believe that humanoid robots can learn to interact with the world by being trained on massive amounts of movement data. Editor’s note: In a recent poll, MIT Technology Review readers selected humanoid robots as the 11th breakthrough for our 2026 list of 10 Breakthrough Technologies. Robotics requires far more complex data about the physical world, though, and that is much harder to find. Virtual simulations can train robots to perform acrobatics, but not how to grasp and move objects, because simulations struggle to model physics with perfect accuracy. For robots to work in factories and serve as housekeepers, real-world data, however time-consuming and expensive to collect, may be what we need. Investors are pouring money feverishly into solving this challenge, spending over $6 billion on humanoid robots in 2025. At-home data recording is becoming a booming gig economy around the world. Data companies like Scale AI and Encord are recruiting their own armies of data recorders, while DoorDash pays delivery drivers to film themselves doing chores. In China, workers in dozens of state-owned robot training centers wear virtual-reality headsets and exoskeletons to teach humanoid robots how to open a microwave and wipe down the table. “There is a lot of demand, and it’s increasing really fast,” says Ali Ansari, CEO of Micro1. He estimates that robotics companies are now spending more than $100 million each year to buy real-world data from his company and others like it. Workers at Micro1 are vetted by an AI agent named Zara that conducts interviews and reviews samples of chore videos. Every week, they submit videos of themselves doing chores around their homes, following a list of instructions about things like keeping their hands visible and moving at natural speed. The videos are reviewed by both AI and a human and are either accepted or rejected. They’re then annotated by AI and a team of hundreds of humans who label the actions in the footage. “There is a lot of demand, and it’s increasing really fast.” Because this approach to training robots is in its infancy, it’s not clear yet what makes good training data. Still, “you need to give lots and lots of variations for the robot to generalize well for basic navigation and manipulation of the world,” says Ansari. But many workers say that creating a variety of “chore content” in their tiny homes is a challenge. Zeus, a scrappy student living in a humble studio, struggles to record anything beyond ironing his clothes every day. Arjun, a tutor in Delhi, India, takes an hour to make a 15-minute video because he spends so much time brainstorming new chores. “How much content [can be made] in the home? How much content?” he says. There’s also the sticky question of privacy. Micro1 asks workers not to show their faces to the camera or reveal personal information such as names, phone numbers, and birth dates. Then it uses AI and human reviewers to remove anything that slips through. But even without faces, the videos capture an intimate slice of workers’ lives: the interiors of their homes, their possessions, their routines. And understanding what kind of personal information they might be recording while they’re busy doing chores on camera can be tricky. Reviews of such footage might not filter out sensitive information beyond the most obvious identifiers. For workers with families, keeping private life off camera is a constant negotiation. Arjun, a father of two daughters, has to wrangle his chaotic two-year-old out of frame. “Sometimes it’s very difficult to work because my daughter is small,” he says. Sasha, a banker turned data recorder in Nigeria, tiptoes around when she hangs her laundry outside in a shared residential compound so she won’t record her neighbors, who watch her in bewilderment. “It’s going to take longer than people think.” While the workers interviewed by MIT Technology Review understand that their data is being used to train robots, none of them know how exactly their data will be used, stored, and shared with third parties, including the robotics companies that Micro1 is selling the data to. For confidentiality reasons, says Ansari, Micro1 doesn’t name its clients or disclose to workers the specific nature of the projects they are contributing to. “It is important that if workers are engaging in this, that they are informed by the companies themselves of the intention … where this kind of technology might go and how that might affect them longer term,” says Yasmine Kotturi, a professor of human-centered computing at the University of Maryland, Baltimore County. Occasionally, some workers say, they’ve seen other workers asking on the company Slack channel if the company could delete their data. Micro1 declined to comment on whether such data is deleted. “People are opting into doing this,” says Ansari. “They could stop the work at any time.” With thousands of workers doing their chores differently in different homes, some roboticists wonder if the data collected from them is reliable enough to train robots safely. “How we conduct our lives in our homes is not always right from a safety point of view,” says Aaron Prather, a roboticist at ASTM International. “If those folks are teaching those bad habits that could lead to an incident, then that’s not good data.” And the sheer volume of data being collected makes reviewing it for quality control challenging. But Ansari says the company rejects videos showing unsafe ways of performing a task, while clumsy movements can be useful to teach robots what not to do. Then there’s the question of how much of this data we need. Micro1 says it has tens of thousands of hours of footage, while Scale AI announced it had gathered more than 100,000 hours. “It’s going to take a long time to get there,” says Ken Goldberg, a roboticist at the University of California, Berkeley. Large language models were trained on text and images that would take a human 100,000 years to read, and humanoid robots may need even more data, because controlling robotic joints is even more complicated than generating text. “It’s going to take longer than people think,” he says. When Dattu, an engineering student living in a bustling tech hub in India, comes home after a full day of classes at his university, he skips dinner and dashes to his tiny balcony, cramped with potted plants and dumbbells. He straps his iPhone to his forehead and records himself folding the same set of clothes over and over again. His family stares at him quizzically. “It’s like some space technology for them,” he says. When he tells his friends about his job, “they just get astounded by the idea that they can get paid by recording chores.” Juggling his university studies with data recording, as well as other data annotation gigs, takes a toll on him. Still, “it feels like you’re doing something different than the whole world,” he says. An exclusive conversation with OpenAI’s chief scientist, Jakub Pachocki, about his firm's new grand challenge and the future of AI. Exclusive: Niantic's AI spinout is training a new world model using 30 billion images of urban landmarks crowdsourced from players. Axiom Math is giving away a powerful new AI tool. But it remains to be seen if it speeds up research as much as the company hopes. One-off tests don’t measure AI’s true impact. We’re better off shifting to more human-centered, context-specific methods. Discover special offers, top stories, upcoming events, and more. Thank you for submitting your email! It looks like something went wrong. We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive. © 2026 MIT Technology Review

In this article Google is adding another robotics partnership to its belt as it leans into robotics as a key bet for artificial intelligence. Agile Robots develops intelligent, sensor-based robotic arms and humanoid robots. The company announced a partnership with Google DeepMind to integrate its Gemini Robotics foundation models with Agile Robots’ hardware. "The partnership is built on a belief that applying AI in the physical world will be transformative," the Tuesday blog post states. "By bringing together Agile Robots' hardware and other AI robotic solutions developed in Germany, with Google DeepMind's Gemini Robotics foundation models, the two teams will improve performance via robot deployment, data collection, model training and iteration." The new partnership means Google will get real-world deployment data as it sees robotics as one of the large use cases for AI, competing against companies like Amazon and Tesla. It also shows the company is making several robotics partnerships as it leans into manufacturing as key use case. Munich-based Agile Robots already has more than 20,000 deployed robotic systems globally and it will integrate Google's tech in existing industrial robots at scale, the blog post says. The partnership will first focus on "high-value industrial" use cases such as manufacturing tasks. "This research partnership is an important step in bringing the impact of AI to the real world," said Carolina Parada, Senior Director and Head of Robotics, Google DeepMind, in Tuesday's blog post. She added that Agile Robots will help Google develop "more advanced AI models for the next generation of robots." In mid-2025, Google debuted two new AI models, Gemini Robotics and Gemini Robotics-ER (extended reasoning), bringing generative AI into physical action commands to control robots. Google said in a blog post at the time that it would partner with Apptronik, a Texas-based robotics developer, to "build the next generation of humanoid robots with Gemini 2.0." In January, Google's DeepMind said it would work with Hyundai's Boston Dynamics, formerly a division of Google, to develop new AI models for its Atlas robot. Last month, Google DeepMind announced that Intrinsic, a robotics software company, will be moved from the "Other Bets" category into the main company with hopes of being "The Android of robotics." The company said it will focus on the manufacturing industry and work with Google's Gemini and infrastructure teams, including potentially helping it with building out Google's own data centers. An early sign that the company was getting serious about robotics was in its hiring of key talent last year. In November, Google's DeepMind unit hired the former CTO of Boston Dynamics Aaron Saunders. However, Google's increased attention to robotics has also brought along internal skepticism. Boston Dynamics, for example, has long-standing contracts with the Defense Department, and some DeepMind employees reportedly brought up concern at an all-hands meeting earlier this year, according to Business Insider. It's not just a trend at Google. Robotics is surfacing as a key use case for AI across the tech industry. In February, Bedrock Robotics, an autonomous vehicle technology startup for construction machinery founded by veterans of Waymo and Segment, raised $270 million in a new fundraising round, valuing the two-year-old start-up at $1.75 billion. The round was led by Alphabet's investment arm CapitalG, Valor Atreides A.I. Fund; Nvidia's venture arm and previous backer 8VC. Got a confidential news tip? We want to hear from you. Sign up for free newsletters and get more CNBC delivered to your inbox Get this delivered to your inbox, and more info about our products and services. © 2026 Versant Media, LLC. All Rights Reserved. A Versant Media Company. Data is a real-time snapshot *Data is delayed at least 15 minutes. Global Business and Financial News, Stock Quotes, and Market Data and Analysis. Data also provided by

The use of AI-powered robots in warfare is no longer just an idea; it is becoming a reality. Modern battlefields are now being used to test advanced machines designed to reduce human risk and improve efficiency. These developments show how quickly robotics and artificial intelligence are moving from labs into real-world situations. One of the most advanced examples is the Phantom MK-1 humanoid robot. It is designed to move like a human and operate in difficult terrains where traditional machines struggle. The robot stands around 175 cm tall, weighs about 80 kg, and can carry up to 20 kg. It uses cameras and sensors to understand its surroundings and can move at speeds of up to 6 km/h. These robots are not fully independent. They are being tested to study mobility, performance, and how AI behaves under pressure. Military robots today use a mix of AI and human control. This is called a “human-in-the-loop” system. AI helps with tasks like identifying objects, navigating terrain, and suggesting actions. However, humans still control critical decisions, especially when it comes to using weapons. Humanoid robots are only part of the story. Uncrewed Ground Vehicles (UGVs) are already widely used. In January 2026 alone, more than 7,000 missions were carried out using robots. These machines mainly handle logistics such as delivering supplies, evacuating injured soldiers, and scouting areas. Most robots are currently used for support tasks rather than direct combat. Despite rapid growth, there are still limitations. Robots face issues like limited battery life, high costs, and difficulty understanding complex situations. There are also concerns about hacking and misuse. Looking ahead, experts believe future warfare could involve large groups of connected robots working together across land, air, and sea. This shift is not just about warfare; it is a major step forward in robotics and AI. Machines are slowly moving from tools to active partners, shaping the future of technology. The Phantom MK-1 is built by a San Francisco-based startup called Foundation, founded by former military personnel and engineers focused on defense robotics. The company has already secured about $24 million in contracts with the US Army, Navy, and Air Force, making it an official defense partner. Beyond this robot, the global race for military robotics is accelerating; countries like the United States, China, Israel, and Russia are actively developing and deploying robotic systems. China has tested armed robot dogs in military drills, while the US has long used systems like PackBot and TALON in combat zones. Even countries like Estonia and Turkey are building advanced unmanned ground and aerial combat systems, showing that the future battlefield is rapidly becoming automated. Read More: (via)

From daily news and career tips to monthly insights on AI, sustainability, Aerospace, and more—pick what matters and get it in your inbox. Access expert insights, exclusive content, and a deeper dive into engineering and innovation. Engineering-inspired textiles, mugs, hats, and thoughtful gifts We connect top engineering talent with the world's most innovative companies. We empower professionals with advanced engineering and tech education to grow careers. We recognize outstanding achievements in engineering, innovation, and technology. All Rights Reserved, IE Media, Inc. Follow Us On Access expert insights, exclusive content, and a deeper dive into engineering and innovation. Engineering-inspired textiles, mugs, hats, and thoughtful gifts We connect top engineering talent with the world's most innovative companies We empower professionals with advanced engineering and tech education to grow careers. We recognize outstanding achievements in engineering, innovation, and technology. All Rights Reserved, IE Media, Inc. The robot will help researchers explore how intelligent systems can better understand and respond to the world around them. Durham University has introduced a new humanoid robot to support advanced research in artificial intelligence, robotics, and human-robot interaction, reflecting a growing trend of universities adopting humanoid platforms for real-world research and experimentation. The university announced this move via its official website on April 1. The robot, named Alan, is a Unitree G1 Edu humanoid that will be used by researchers and students as a shared research platform to explore how robots can operate alongside humans, perform complex tasks, and function autonomously in dynamic environments. The platform is designed specifically for education and research institutions, allowing universities to experiment with artificial intelligence and robotics software on a full humanoid system. Humanoid robots are particularly valuable in research because they are designed to operate in environments built for humans. This allows researchers to test robots in realistic settings, such as laboratories, offices, and public spaces, without specialized infrastructure. The Unitree G1 platform has been used in a variety of robotics research and demonstrations, showcasing capabilities such as autonomous walking, playing games, interacting with objects, and performing complex movement tasks. The Unitree G1 has 23 degrees of freedom and full-body mobility, enabling it to perform tasks that require balance, manipulation, and coordination. Alan will primarily serve as a shared research platform within Durham University’s Computer Science department, particularly supporting the work of the VIViD research group. Researchers plan to use the humanoid robot to study how robots can recognize people and objects, understand complex scenes, imitate human actions, and make decisions in everyday environments. The platform will enable researchers to explore how intelligent robotic systems perceive and interact with their environments. The robot may also support research in assistive robotics, an area focused on developing robots that can work safely and usefully alongside people in real-world settings. This includes exploring how robots could assist humans in daily activities while operating safely in shared environments. In addition to these areas, the Unitree G1 will contribute to broader research projects across the department. Research involving humanoid robots has expanded rapidly in recent years, with platforms like the Unitree G1 used in experiments spanning sports and motion learning to industrial automation and autonomous navigation. One of the next research areas is exploring how the robot can perform simple tasks and make real-time decisions without relying heavily on external computing support. This activity would allow the humanoid robot to operate more independently and function more effectively in real-world environments. As a physical research platform, the robot enables researchers to test ideas in a practical, controlled way rather than solely through simulations or software models. The Unitree G1 will also support the department’s ongoing work in artificial intelligence, robotics, and visual computing, while providing opportunities for collaboration across research groups and projects. Atharva is a full-time content writer with a post-graduate degree in media & amp; entertainment and a graduate degree in electronics & telecommunications. He has written in the sports and technology domains respectively. In his leisure time, Atharva loves learning about digital marketing and watching soccer matches. His main goal behind joining Interesting Engineering is to learn more about how the recent technological advancements are helping human beings on both societal and individual levels in their daily lives. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Exclusive content, expert insights and a deeper dive into engineering and tech. No ads, no limits. Premium Follow

Deep Reinforcement Learning (RL) is now commonly used for controlling legged robots. Several recent studies have demonstrated impressive results in solving increasingly complex robotic tasks such as navigation in unstructured environments or loco-manipulation. However, this complexity often comes with intricate learning setups requiring tedious reward shaping and features to help convergence. In this work, we tackle these issues and achieve loco-manipulation with a humanoid robot using a RL algorithm that enforces constraints through stochastic terminations during policy learning. We keep the number of rewards low by reformulating them as constraints when they can be intuitively expressed that way. Moreover, we study the relevance of various learning features encountered in the literature and show that providing observations without noise or privileged information to the critic are two straightforward ways to boost locomotion performances on rough terrains. We also demonstrate that the proposed minimalist architecture is not limited to pure locomotion but extends to a loco-manipulation task involving upper limbs. Videos are available at humanoid-cat.github.io. Connectez-vous pour contacter le contributeur https://hal.science/hal-05553678 Soumis le : lundi 16 mars 2026-08:47:28 Dernière modification le : mardi 17 mars 2026-03:18:45 Contact Ressources Informations Questions juridiques Portails CCSD

Posts from this topic will be added to your daily email digest and your homepage feed. See All Tech Posts from this topic will be added to your daily email digest and your homepage feed. See All AI Posts from this topic will be added to your daily email digest and your homepage feed. See All News AI can help develop methods of locomotion that are unconventional but fast AI can help develop methods of locomotion that are unconventional but fast Posts from this author will be added to your daily email digest and your homepage feed. See All by James Vincent Quadrupedal robots are becoming a familiar sight, but engineers are still working out the full capabilities of these machines. Now, a group of researchers from MIT says one way to improve their functionality might be to use AI to help teach the bots how to walk and run. Usually, when engineers are creating the software that controls the movement of legged robots, they write a set of rules about how the machine should respond to certain inputs. So, if a robot’s sensors detect x amount of force on leg y, it will respond by powering up motor a to exert torque b, and so on. Coding these parameters is complicated and time-consuming, but it gives researchers precise and predictable control over the robots. AI uses trial and error to develop its own style of running An alternative approach is to use machine learning — specifically, a method known as reinforcement learning that functions through trial and error. This works by giving your AI model a goal known as a “reward function” (e.g., move as fast as you can) and then letting it loose to work out how to achieve that outcome from scratch. This takes a long time, but it helps if you let the AI experiment in a virtual environment where you can speed up time. It’s why reinforcement learning, or RL, is a popular way to develop AI that plays video games. This is the technique that MIT’s engineers used, creating new software (known as a “controller”) for the university’s research quadruped, Mini Cheetah. Using reinforcement learning, they were able to achieve a new top-speed for the robot of 3.9m/s, or roughly 8.7mph. You can watch what that looks like in the video below: As you can see, Mini Cheetah’s new running gait is a little ungainly. In fact, it looks like a puppy scrabbling to accelerate on a wooden floor. But, according to MIT PhD student Gabriel Margolis (a co-author of the research along with postdoc fellow Ge Yang), this is because the AI isn’t optimizing for anything but speed. “RL finds one way to run fast, but given an underspecified reward function, it has no reason to prefer a gait that is ‘natural-looking’ or preferred by humans,” Margolis tells The Verge over email. He says the model could certainly be instructed to develop a more flowing form of locomotion, but the whole point of the endeavor is to optimize for speed alone. “Programming how a robot should act in every possible situation is simply very hard” Margolis and Yang say a big advantage of developing controller software using AI is that it’s less time-consuming than messing about with all the physics. “Programming how a robot should act in every possible situation is simply very hard. The process is tedious because if a robot were to fail on a particular terrain, a human engineer would need to identify the cause of failure and manually adapt the robot controller,” they say. By using a simulator, engineers can place the robot in any number of virtual environments — from solid pavement to slippery rubble — and let it work things out for itself. Indeed, the MIT group says its simulator was able to speed through 100 days’ worth of staggering, walking, and running in just three hours of real time. Some companies that develop legged robots are already using these sorts of methods to design new controllers. Others, though, like Boston Dynamics, apparently rely on more traditional approaches. (This makes sense given the company’s interest in developing very specific movements — like the jumps, vaults, and flips seen in its choreographed videos.) There are also faster-legged robots out there. Boston Dynamics’ Cheetah bot currently holds the record for a quadruped, reaching speeds of 28.3 mph — faster than Usain Bolt. However, not only is Cheetah a much bigger and more powerful machine than MIT’s Mini Cheetah, but it achieved its record running on a treadmill and mounted to a lever for stability. Without these advantages, maybe AI would give the machine a run for its money. Posts from this author will be added to your daily email digest and your homepage feed. See All by James Vincent Posts from this topic will be added to your daily email digest and your homepage feed. See All AI Posts from this topic will be added to your daily email digest and your homepage feed. See All News Posts from this topic will be added to your daily email digest and your homepage feed. See All Robot Posts from this topic will be added to your daily email digest and your homepage feed. See All Science Posts from this topic will be added to your daily email digest and your homepage feed. See All Tech A free daily digest of the news that matters most. This is the title for the native ad This is the title for the native ad © 2026 Vox Media, LLC. All Rights Reserved

If you want to work with robots you can do all sorts of learning with software and simulation, but nothing quite beats getting to grips with real machinery. That was the motivation for [James Gullberg] to build this impressive robot arm. Featuring six degrees of freedom, the robot arm is mostly constructed of 3D printed components. This let [James] experiment with a wide variety of joint and reducer designs for the sake of learning and investigation. The base of the robot uses a fairly conventional planetary gear drive, while shoulder and elbow joints rely on split-ring planetary gearboxes to allow for high torque density with regards to size. [James] implemented a neat sensing technique here, integrating alternating magnets into the output ring gear which are monitored via a magnetic encoder. The wrist joint switches things up again, running via an inverted belt differential. Running the show is an STM32 microcontroller, which talks to all the encoders, communicates with a Raspberry Pi over CAN bus, and handles all the necessary PID control loops and step generation for the drive motors. The plan is to run higher-level control on the Raspberry Pi which will run a ROS 2-based software stack. Already, the various joints look smooth and impressive in motion. If you’re looking to learn about robot arms, you really can’t beat building one. We’ve featured a few projects along these lines before. Most of them aren’t exactly production-line ready, but they will teach you a ton about control, motion planning, and all sorts of associated skills. That experience can be invaluable if you intend to work with robots in industry. My (mostly) 3D printed Robot Arm byu/SPACE-DRAGON772 inEngineeringPorn Thanks to [JohnU] for the tip! Wow! Looks great! I wonder what kind of load those gears can handle. Either way I want one. Maybe I could finally get a robot to do the dishes for me. I am quite curious about how this thing really moves. From the 3D printed gears, my first guess is that there will be quite a lot of backlash, and this seems to be “cleverly” hidden in the video by only showing short clips of moves at constant speed. And even then you can see a bit of jerkiness. I am also having some doubts about the carbon fiber tubes. On themselves such tubes are extremely stiff, but with such long and thing tubes, any movement or flex in the plastic part where they are mounted will be amplified greatly. One of the better tests for mechanical stability is to just take the end effector by the hand and see how much it moves when you pull and push it a bit. Or with quick start and stop motions. Backlash and flex are very common causes that reduces good looking robot arms to not much more then “demo’s”. Look like there is a timing belt, and a two stage planetary gear in the shoulder joint and a total gear ratio of around 1:30 and that is a good compromise between force, speed and resolution. Another aspect I like are the arc of magnets. These are almost certainly part of an angle measurement system, and when done properly this can have a quite decent resolution. Use of ball bearings is also nice. I also like the use of standard nuts in a lot of parts. This is probably stronger then threaded inserts. In some places this could maybe still be improved upon a bit by using longer bolts and putting the nuts “on the other side” so the plastic is only used in compression. But overall, this design has quite a lot going for it. On his website he mentions that more info will follow soon. So I’ll save a link and come back later. Wondering how this compares to the Anin AR4 ? Was thinking of getting one. Please be kind and respectful to help make the comments section excellent. (Comment Policy) This site uses Akismet to reduce spam. Learn how your comment data is processed.

Multi-robot systems represent a key class of multi-agent systems where multiple autonomous agents cooperate to achieve tasks beyond the capabilities of a single robot. Their effectiveness relies on decentralized or distributed control, where collective behaviors emerge from local interactions under limited information and communication. Classical approaches have enabled significant progress in understanding coordination, information flow, and connectivity maintenance. However, these methods often depend on strong modeling assumptions and struggle to scale in dynamic, uncertain environments. Recent advances in machine learning provide promising alternatives by allowing agents to learn coordination strategies from data, enabling robustness and adaptability under partial observability and noisy sensing. Yet, existing learning-based frameworks rarely account for explicit communication, a critical factor in scalable multi-robot coordination that has been central to control-theoretic approaches. This dissertation addresses this gap by developing hybrid methods that integrate learning with communication-aware distributed control. By combining the generalization and flexibility of machine learning with the theoretical guarantees of control theory, this work advances the foundations of collective decision-making and contributes to the deployment of communication-aware, learning-based multi-robot systems in real-world applications. Les systèmes multi-robots constituent une classe centrale de systèmes multi-agents, où plusieurs robots coopèrent pour accomplir des tâches dépassant les capacités d’un seul agent. Leur efficacité repose sur des mécanismes décentralisés ou distribués, mais les approches classiques, bien qu’efficaces pour analyser la coordination et le maintien de la connectivité, peinent à s’adapter à des environnements dynamiques et incertains. L’apprentissage automatique offre une alternative prometteuse en permettant aux agents d’apprendre des stratégies de coordination robustes à partir de données, mais il intègre rarement la communication explicite, pourtant essentielle à l’évolutivité. Cette thèse propose des méthodes hybrides combinant apprentissage et contrôle distribué sensible à la communication, afin de concevoir des systèmes multi-robots plus adaptatifs, robustes et déployables dans des environnements réels. Contact https://theses.hal.science/tel-05549372 Soumis le : jeudi 12 mars 2026-15:02:05 Dernière modification le : jeudi 9 avril 2026-11:20:26 Contact Ressources Informations Questions juridiques Portails CCSD

Oops, something went wrong NXP Semiconductors N.V. (NASDAQ:NXPI) is one of the 7 Cheapest AI Data Center Stocks to Buy Now. On March 17, 2026, NXP Semiconductors N.V. (NASDAQ:NXPI) reported robotics solutions jointly with Nvidia, combining Nvidia Holoscan Sensor Bridge with NXP SoCs for sensor fusion, machine vision, and motor control. The corporation noted savings in component size, footprint, power, and cost savings while focusing on applications in humanoid robotics and physical AI systems. The corporation also outlined its efforts in edge processing, secure networking, and real-time control for robotics systems. NXP Semiconductors N.V. (NASDAQ:NXPI) reported fourth-quarter revenue of $3.34 billion, up 7% year on year, and full-year revenue of $12.27 billion, a 3% decrease year on year. It posted non-GAAP diluted EPS of $3.35 in the fourth quarter and $11.81 in the full year, with $793 million in non-GAAP free cash flow in the quarter. It also bought back $338 million in shares and paid out $254 million in dividends during the quarter, with an additional $36 million in repurchases conducted after the quarter ended. A semiconductor. Photo by Tima Miroshnichenko on Pexels NXP Semiconductors N.V. (NASDAQ:NXPI) is a holding company that provides semiconductor solutions. It operates in the following areas: China, the Netherlands, the United States, Singapore, Germany, Japan, South Korea, Malaysia, and other countries. While we acknowledge the potential of NXPI as an investment, we believe certain AI stocks offer greater upside potential and carry less downside risk. If you’re looking for an extremely undervalued AI stock that also stands to benefit significantly from Trump-era tariffs and the onshoring trend, see our free report on the best short-term AI stock. READ NEXT: 33 Stocks That Should Double in 3 Years and Cathie Wood 2026 Portfolio: 10 Best Stocks to Buy. Disclosure: None. Follow Insider Monkey on Google News. Sign in to access your portfolio

Unitree Robotics is set to launch its cheapest humanoid robot, the R1, in global markets next week. What makes this launch stand out is the sales strategy. Unitree is taking humanoid robots directly to online platforms, starting with AliExpress. This marks a shift from traditional enterprise sales to a more accessible, e-commerce-driven approach. The initial rollout will cover North America, Europe, Japan, and Singapore. The company is also joining Alibaba’s Brand+ channel, which brings benefits like free shipping, free returns, and better global visibility. More online and offline sales channels are expected to follow. The R1 was first launched in China in 2025 at a starting price of 29,900 yuan (around $4,370), making it one of the most affordable humanoid robots available. While global pricing is still unannounced, the company is expected to keep it competitive. Standing 123 cm tall and weighing over 27 kg, the R1 is designed for dynamic movement. It can perform cartwheels, run downhill, stand up from the ground, and execute complex routines. The “Born for sport” positioning highlights its focus on agility and motion rather than industrial tasks. Unitree’s move to sell robots online signals a major shift in how humanoid machines are marketed and distributed. Until now, most robots were sold through enterprise deals or research contracts. By listing on platforms like AliExpress, Unitree is opening access to developers, educators, and even early consumers globally. This builds on its existing customer base, where around 70% of shipments in 2025 went to universities and research institutions. The online model could further expand this ecosystem. The company shipped over 5,500 humanoid robots in 2025, far ahead of competitors like Tesla, Figure AI, and Agility Robotics, which delivered around 150 units each. Unitree now aims to ship 10,000 to 20,000 units in 2026. Its cost advantage comes from a highly localized supply chain, with over 80% of components sourced within China. This allows it to price robots far below global averages, where similar machines can cost up to $300,000. Unitree’s online-first global launch could reshape the humanoid robotics market. By combining low pricing with e-commerce accessibility, the company is testing real-world demand beyond labs and factories. If successful, this approach could accelerate adoption and bring humanoid robots closer to mainstream buyers much faster than expected. Read More:

UBTech’s salary range for Chief Scientist of Embodied Intelligence runs from $2.2M to $18M. The Shenzhen company’s humanoid robot revenue grew twenty-fold last year. Bloomberg calls the offer unusual even by Chinese standards. Chinese humanoid robotics company UBTech has posted a global recruitment notice for a Chief Scientist of Embodied Intelligence, offering an annual salary ranging from 15 million to 124 million yuan, equivalent to $2.2 million to $18 million. The role, which the company’s WeChat post describes as setting UBTech’s full technology roadmap and being ‘the helmsman of UBTECH’s technical path,’ was confirmed to China’s Global Times. Bloomberg described the package as unusual even by Chinese standards, noting that China’s AI industry has historically avoided the mega compensation packages commonplace in Silicon Valley. TNW City Coworking space - Where your best work happens A workspace designed for growth, collaboration, and endless networking opportunities in the heart of tech. UBTech was founded in 2012 and is headquartered in Shenzhen. It became the world’s first publicly listed humanoid robot maker, trading in Hong Kong. Its primary product is the Walker S2, a 5-foot-9 humanoid designed to operate autonomously in factories. Earlier this year, UBTech struck a deal with Airbus to test Walker S2 units on aircraft manufacturing production lines. In its annual results released on 31 March, the company reported 2025 revenue of 2.01 billion yuan, up 53.3% year-on-year. Revenue from humanoid products and services specifically reached 820.6 million yuan, representing 41% of total revenue, a twenty-fold increase from the previous year when that segment generated just 35.6 million yuan and accounted for less than 3% of revenue. The Chief Scientist role will lead research in vision-language-action models, robotics foundation models, and manipulation and dexterity capabilities, with the stated goal of accelerating large-scale deployment across manufacturing, commercial services, and what UBTech describes as ‘family companionship.’ The package is structured as a combination of cash, benefits, and equity. The job posting, which states UBTech does not care about passports, age, or gender, and asks only ‘Can you define the future?’, is part of a broader hiring push that also includes reinforcement learning algorithm engineers, hardware engineers, and EtherCAT master system developers. The offer lands at a moment when China’s humanoid robot industry is receiving explicit government support. Premier Li Qiang has included robotics in the government work report for two consecutive years, and Chinese companies accounted for nearly 90% of global humanoid robot shipments in 2025, according to research firm Omdia. UBTech sold 1,079 full-size humanoid robots last year, generating 820 million yuan. Tesla, meanwhile, posted a notice in late March seeking more than 80 specialists for its Optimus humanoid programme. The AI talent war that began in large language models is moving rapidly into embodied intelligence. I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, (show all) I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, and change. I write about startups, venture capital, digital policy, and Europe ecosystem, with an eye on the larger story beneath them: who gets to build the future, who profits from it, and how Europe is learning to speak in a louder voice of its own. Before moving into senior editorial leadership, I've built my career for over +10 years across journalism, storytelling, content strategy, SEO, and digital publishing, with experience in SaaS, hospitality, art, and culture. Get the most important tech news in your inbox each week. The heart of tech A Tekpon Company Copyright © 2006—2026, Cogneve, INC. Made with <3 in Amsterdam.

Den Delamarski is a Principal Product Engineer at Microsoft working in the Core AI division, focusing on AI tools for developers. Den is one of the core maintainers of the Model Context Protocol (MCP), having initially joined the project through his expertise in security and authorization. When MCP first launched with an auth specification, Den identified opportunities to improve it for enterprise scale and worked with the Anthropic team to rewrite the authorization framework, which was merged into the June 2024 version of the protocol. Beyond MCP, Den helps drive projects like GitHub SpecKit, which enables spec-driven development with agentic coding tools. His work centers on building developer tools and experiences in the rapidly evolving AI ecosystem, including projects like Copilot and other Microsoft AI initiatives. The Model Context Protocol solves a fundamental problem in AI systems: LLMs are trained on data that gets locked at a specific point in time, but users need to work with fresh, dynamic data. MCP provides a universal interface that allows any LLM or AI client to connect to data sources, applications, and services without custom integrations. Just as the Language Server Protocol (LSP) standardized how editors communicate with language tools, MCP standardizes how AI agents access external capabilities. The protocol is highly opinionated about authentication, message passing, and primitive exposure, eliminating the inconsistency found in traditional REST API integrations. The protocol went from non-existent to widely adopted in less than a year, with major companies across banking, healthcare, and gaming now integrating MCP into their AI strategies. The composability of MCP means you can connect multiple servers to a single client, allowing an LLM to coordinate across different data sources and services seamlessly. The Python SDK makes building MCP servers remarkably simple through the FastMCP framework, which provides a Flask-like developer experience. Creating an MCP tool is as straightforward as writing a Python function and adding a decorator. The SDK handles all the complex JSON-RPC envelope creation, streaming, and protocol compliance automatically. Developers can focus on business logic rather than protocol implementation details. FastMCP is integral to the official Python SDK and simplifies common pain points like authorization. The programming model supports async functions naturally, allowing you to await user input via elicitations without complex callback patterns. The framework also includes built-in support for structured output using Pydantic models, progress reporting, and image handling. MCP servers expose three fundamental primitives that LLMs can interact with. Tools are function calls that perform actions - think of them as API endpoints that do something like sending an email, querying a database, or creating a 3D scene in Blender. Prompts are reusable templates that help LLMs interact with your server effectively, such as "create a recipe with substitutions." Resources allow LLMs to reference and work with entities like databases, files, or API objects. Each primitive serves a distinct purpose in the agent workflow. Tools enable actions and side effects. Prompts guide the LLM on how to best use your server. Resources provide structured access to data and entities. Together, these primitives create a complete interaction model that's both powerful and constrained enough to be reliable. MCP servers can run in two distinct modes depending on your architecture needs. Local MCP servers use stdio (standard input/output) to communicate via native OS constructs between the MCP client and server processes. This is perfect for development machines where you want your editor or AI tool to access local capabilities without network overhead. Remote MCP servers use streamable HTTP and can be hosted anywhere - AWS, Azure, GCP, your home lab, or behind a reverse proxy like Nginx or Caddy. The transport layer is abstracted by the SDK, so the same server code can work in both modes with minimal changes. For local development with remote access, tools like Tailscale provide secure overlay networks without exposing ports or configuring complex VPN setups. This makes it trivial to run an MCP server on your home lab and access it securely from anywhere. The MCP Registry launched in September 2024 as a centralized API that indexes all publicly available MCP servers. Think of it like Docker Hub for MCP servers - you can discover servers, but you're not required to use the registry. The registry supports both public servers (like the GitHub-maintained registry) and private enterprise registries for internal company use. This allows organizations to maintain approved MCP servers behind security gates while still benefiting from the discoverability infrastructure. Discovery is improving rapidly with better integration into clients like VS Code, Cursor, and Claude Desktop. The Awesome MCP Servers list on GitHub has become a valuable community resource with hundreds of servers categorized by function - from biology and medicine to gaming, marketing, and sports analytics. Security and authorization was Den's entry point into MCP development. The June 2024 spec introduced formal OAuth 2.1-based authorization, eliminating the need for developers to implement custom auth flows or check API keys into source control. The brilliant part is that MCP server developers don't need to become OAuth experts - the SDKs handle it. For consumers, authentication is as simple as logging in when you connect a server. The client bootstraps the auth flow, stores tokens securely, and ensures you access only your data. MCP servers can specify whether they use API keys (stored in configuration) or OAuth (handled via standard browser-based login flows). This approach scales from hobby projects to enterprise deployments where data access controls are critical. The standardization means you don't face "17 different dances" to get authentication tokens from different services. GitHub SpecKit represents Microsoft's hypothesis for how spec-driven development works with AI coding tools. The approach starts with defining what and why you're building in a specification document, then breaks down the technical implementation plan, and finally decomposes it into consumable tasks that AI can execute iteratively or in parallel. This isn't the only way to do spec-driven development, but it provides a recipe book and ingredient box for teams wanting to adopt this workflow. The philosophy recognizes that there's no single correct approach to spec-driven development - it depends on your models, team structure, and project complexity. However, starting with a thorough planning phase using high-quality models, then executing with faster models guided by those specs, has proven effective for managing AI agent workflows on complex projects. The MCP ecosystem has exploded with creative and practical implementations. The Blender MCP server lets you describe a medieval scene with a dragon and lighting, and it builds it for you by translating high-level descriptions into Blender's native API calls. Gaming servers exist for Unity 3D, Minecraft, and even analyzing Halo stats. Marketing professionals can connect Facebook Ads, Google Ads, and Amazon Ads MCP servers to ask "how are my ads performing this week" across all platforms without clicking through dashboards. Sports enthusiasts can use Strava MCP for running and biking analytics, or the Formula 1 Multiviewer MCP that controls viewing angles and telemetry during live races. For developers, there are Jira and Atlassian MCP servers to automate bug triage and ticket management. The diversity shows MCP's flexibility - it's not just for data retrieval, but for controlling applications, analyzing information, and automating workflows across domains. Retrieval Augmented Generation (RAG) and MCP serve different purposes in the AI architecture landscape. RAG builds vector databases to augment an LLM's context with additional knowledge, helping it understand what exists in a codebase or documentation set. It's primarily about giving the LLM more relevant context for making decisions. MCP, on the other hand, provides universal access to live data and actionable capabilities. It's not just about knowing what exists - it's about doing something with that information. While RAG helps an LLM understand that an authorization component exists in your codebase, MCP lets it actually invoke authentication services, update records, or chain multiple actions across services. The two technologies can complement each other: RAG for knowledge augmentation and MCP for capability extension. Many real-world AI applications benefit from using both - RAG for understanding context and MCP for taking action. There's ongoing debate about whether specialized local models or general-purpose cloud models work better for specific tasks. Den's perspective is that general-purpose models like Claude and GPT-4 will typically outperform local models for most scenarios due to superior training resources and compute power. However, local models excel for privacy-sensitive workloads - like organizing family photos without sending them to remote servers - or domain-specific tasks where a small, focused model can be as effective as a large general one. MCP enables an interesting hybrid approach: use powerful general-purpose models for orchestration and decision-making, but delegate specific subtasks to specialized local models or services via MCP servers. For example, a general model could coordinate a photo organizing workflow while a local computer vision model handles the actual image analysis. This composability allows building sophisticated systems that balance capability, privacy, cost, and latency. The Python MCP SDK prioritizes developer experience through familiar patterns and minimal boilerplate. The decorator-based approach (@mcp.tool) mirrors Flask and FastAPI, making it immediately intuitive for Python web developers. Async/await support is first-class, allowing natural progress reporting and elicitations without callback hell. The SDK includes 143+ contributors, ships releases every few days, and maintains "good first issue" tags for new contributors. Documentation and samples are comprehensive, with the official Python SDK repo containing multiple example servers. The team actively solicits feedback and iterates quickly on developer pain points. Installation is as simple as uv add mcp or pip install mcp, and you can have a working MCP server in under 10 lines of code. The combination of low barrier to entry and production-ready features makes MCP accessible to Python developers at all skill levels. While MCP provides secure authentication mechanisms, users must still exercise caution when installing third-party MCP servers. Like any software that accesses your data, you should verify the source and understand what an MCP server does before connecting it. An MCP server that reads your iMessages to "sort by importance" could potentially scan for credit card numbers or social security numbers. The responsibility for vetting servers lies with the user, just as it does with browser extensions or system-level applications. Best practices include reviewing source code for open-source MCP servers, starting with servers from trusted organizations, using private registries for enterprise deployments, and being cautious about granting broad permissions. Never check API keys into source control - use environment variables or OAuth flows instead. The MCP community is working on improved discovery with trust signals, but individual diligence remains essential for security. "Think about it like last year at this time, like at the time when we were recording the work item episode, MCP did not exist. Just not a thing. And now everybody's talking about MCP. Like you talk to any big companies, you talk to like the banks, the healthcare, the gaming, everybody, everybody cares about MCP." -- Den Delamarski "The way the folks at Anthropic have been describing it, it is USB-C for AI." -- Den Delamarski "Look at the simplicity of this. You literally have a Python function, you have def add, and there is your arguments, you would pass you a function, like two integers. And then all you need to do to make that a tool that an LM can invoke is just add that @mcp.tool decorator. That's it. You're not going and crafting elaborate JSON RPC envelopes and converters and all these things." -- Den Delamarski on the developer experience "I'll tell you what, the LLMs are getting really good at analyzing the stats. You give them the data, they can make some conclusions." -- Den Delamarski on his Halo stats MCP server "Do you remember the days when you had to work, this episode is not sponsored by Tailscale, for the record. Should be." -- Den Delamarski and Michael Kennedy discussing VPN complexity vs. Tailscale simplicity "The power is composability. It's the fact that you can compose things together and have them work together based on the prompts that you have and scenarios that you have." -- Den Delamarski "There's an MCP server for everything. Like, this list is massive. I'm actually like, every time I discover these things, I was like, oh, I didn't know there was one for multiviewer." -- Den Delamarski exploring the Awesome MCP Servers list "These are the life hacks you learned only from this podcast. Query all the bugs assigned to me, reassign them to somebody else." -- Den Delamarski joking about Jira MCP automation "Exercise caution, just like you would exercise with any other software and APIs and websites where you log in because the responsibility is kind of on you to figure out what's safe, what's not." -- Den Delamarski on MCP server security Model Context Protocol (MCP): An open protocol that provides a standardized way for AI applications to connect to data sources, services, and tools. It acts as a universal translation layer between LLMs and external systems, similar to how LSP standardized language tooling for editors. MCP Server: A service that implements the MCP specification and exposes tools, prompts, and resources that AI clients can use. Servers can run locally via stdio or remotely via HTTP. MCP Client: An application or editor that connects to MCP servers and makes their capabilities available to LLMs. Examples include VS Code, Cursor, Claude Desktop, and custom applications. Tools: Function calls that MCP servers expose to LLMs, allowing them to perform actions like querying databases, sending emails, or controlling applications. Prompts: Reusable templates that MCP servers provide to guide LLMs on how to interact effectively with their capabilities. Resources: References to databases, files, or API entities that MCP servers make available to LLMs for data access and manipulation. Elicitations: A mechanism for MCP servers to request structured input from users during tool execution, enabling confirmation dialogs, dropdown selections, and data validation. FastMCP: The primary framework within the Python SDK that provides a Flask-like decorator-based programming model for building MCP servers quickly. stdio Transport: A local communication method where MCP servers use standard input/output pipes to exchange JSON-RPC messages with clients on the same machine. Streamable HTTP Transport: A remote communication method where MCP servers expose HTTP endpoints for JSON-RPC message exchange, enabling cloud deployment and distributed architectures. JSON-RPC: The underlying message format used by MCP for communication between clients and servers, abstracted away by SDKs for developer convenience. MCP Registry: A centralized index of available MCP servers, similar to Docker Hub, that enables discovery and installation of servers into MCP clients. Supports both public and private registries. OAuth 2.1: The authentication and authorization standard used by MCP for secure access to protected resources, handled automatically by SDKs. RAG (Retrieval Augmented Generation): A technique that builds vector databases to augment LLM context with additional knowledge, complementary to MCP's action-oriented approach. Spec-Driven Development: A development methodology where projects start with detailed specifications that guide AI coding tools through implementation, promoted by GitHub SpecKit. If you want to dive deeper into the topics covered in this episode, these courses from Talk Python Training can help you build the foundational skills and advanced techniques you'll need. LLM Building Blocks for Python: This concise 1.2-hour course teaches you to move beyond basic "text in, text out" with LLMs, covering structured data, chat workflows, async pipelines, and caching - essential skills for building MCP servers that integrate AI capabilities. Modern APIs with FastAPI and Python: Since FastMCP uses FastAPI-like patterns, this course provides deep knowledge of building modern Python APIs with type hints, async/await, and data validation - all of which directly apply to MCP server development. Async Techniques and Examples in Python: MCP servers heavily use async/await for streaming responses and progress reporting. This course covers Python's entire async ecosystem, from basic async/await to parallel processing and thread safety. Rock Solid Python with Python Typing: Type hints are fundamental to MCP servers and structured output with Pydantic. Learn how to use Python's typing system effectively, which powers frameworks like FastAPI and FastMCP. Build An Audio AI App: This course combines AI, FastAPI, and MongoDB to build real applications - a perfect companion for creating MCP servers that work with audio content, transcripts, and multimedia data. The Model Context Protocol represents a fundamental shift in how we build AI-powered applications. Rather than creating custom integrations for every data source and service, MCP provides a universal standard that works across LLMs, editors, and agentic tools. The Python ecosystem has embraced MCP with remarkable speed, delivering a developer experience that feels as natural as Flask or FastAPI while handling the complexity of JSON-RPC, streaming, and authentication behind the scenes. What makes MCP truly powerful is its composability. You can connect multiple servers to a single client, enabling LLMs to coordinate sophisticated workflows across different services. The registry ecosystem is exploding with servers for everything from 3D modeling in Blender to analyzing Formula 1 telemetry to automating Jira tickets. Yet beneath this diversity lies a consistent, well-designed protocol that makes both building and consuming MCP servers straightforward. For Python developers, now is the perfect time to explore MCP. The barriers to entry are low - you can have a working server in minutes. The community is active and welcoming, with good first issues available for contributors. The use cases span every domain imaginable, from enterprise data integration to creative hobby projects. Whether you're building the next generation of AI agents or simply want to give your AI tools access to your custom data, MCP provides the plumbing that just works. As Den put it, "MCP can do anything - it's just a pipe. What you do with that pipe is up to you." 00:00 On this episode, we're digging into the Model Context Protocol, or MCP. 00:04 Think LSP for AI. Build a small Python service once, and your tools and data show up across 00:11 editors and agents like VS Code, Claude Code, and more. My guest, Den Delamarski from Microsoft, 00:17 helps build this space and keeps us honest about what's solid versus what's just shiny. 00:23 We'll keep it practical, transports that actually work, guardrails you can trust, 00:27 and a tiny server you could ship this week. 00:29 By the end, you'll have a clear mental model and a path to plug Python into the internet of agents. 00:36 This is Talk Python To Me, episode 526, recorded September 30th, 2025. 00:43 Talk Python To Me, yeah, we ready to roll. 00:46 Upgrading the code, no fear of getting old. 00:48 Async in the air, new frameworks in sight. 00:51 Geeky rap on deck. 00:52 Quarth crew, it's time to unite. 00:54 We started in Pyramid, cruising old school. 00:57 lanes. Had that stable base. Yes. Welcome to Talk Python To Me, the number one podcast for Python 01:02 developers and data scientists. This is your host, Michael Kennedy. I'm a PSF fellow who's been coding 01:07 for over 25 years. Let's connect on social media. You'll find me and Talk Python on Mastodon, 01:13 Bluesky, and X. The social links are all in the show notes. You can find over 10 years of past 01:19 episodes at talkpython.fm. And if you want to be part of the show, you can join our recording 01:24 live streams. That's right. We live stream the raw uncut version of each episode on YouTube. 01:30 Just visit talkpython.fm/youtube to see the schedule of upcoming events. And be sure to 01:36 subscribe and press the bell so you'll get notified anytime we're recording. This episode is sponsored 01:41 by Posit Connect from the makers of Shiny. Publish, share, and deploy all of your data projects that 01:47 you're creating using Python. Streamlit, Dash, Shiny, Bokeh, FastAPI, Flask, Quarto, Reports, 01:54 dashboards, and APIs. Posit Connect supports all of them. Try Posit Connect for free by going to 02:00 talkpython.fm/posit, P-O-S-I-T. And it's brought to you by Nordstellar. Nordstellar is a 02:07 threat exposure management platform from the Nord security family, the folks behind NordVPN, 02:13 that combines dark web intelligence, session hijacking prevention, brand and domain abuse 02:19 detection, and external attack surface management. Learn more and get started keeping your team safe 02:24 at talkpython.fm/nordstellar. Hey, I want to take just a minute and talk to you guys. I just 02:31 released a really cool new course called Agentic AI Programming for Python Developers and Data 02:36 Scientists. You've heard me mention a couple times on the podcast how I've had some incredible success 02:42 with some of these Agentic AI coding tools. I hear people talking about how they're not really 02:47 working for them. And then I look at the results that I'm getting and think, wow, that's something 02:53 that would have taken two weeks. It's built in two hours and it's well factored and good looking code. 03:00 What gives? Why is this difference here? Well, I decided to create this course to share all the 03:06 things that I'm doing with these agentic coding tools with the idea of making you as successful 03:12 and productive as well. Yes, I know we're all tired about hearing about how AI is going to 03:17 change everything for software developers. 03:19 But there are some tools here that will give you truly difference 03:23 making levels of productivity. 03:25 And that's what this course is about. 03:27 So check it out at talkpython.fm/agenticai. 03:31 The links in your podcast player show notes. 03:33 Let's get to the interview. 03:35 Ben, welcome to Talk Python To Me. 03:36 Great to have you here. 03:37 Hello, hello. 03:38 I'm excited to be here. 03:39 I'm a big fan of Talk Python. 03:41 I'm a big fan of you and I'm a big fan of Python. 03:43 So there we go. 03:45 Wow. 03:45 Thank you. 03:46 I've been on your show, Work Item, which was really fun. 03:49 Thank you for having me. 03:50 And now it's time to dive into your expertise. 03:53 I'm going to talk agentic stuff, and especially we're going to talk model context protocol, MCP. 04:01 I think this is one of the really important layers that is kind of invisible, right? 04:05 A lot of the coding agents and coding AI and chat LLMs and all that, 04:10 that's what people think when they hear all these things. 04:13 But there's got to be plumbing, right? 04:15 We're going to talk to plumbing. 04:16 - There has to be, yeah. 04:17 - Nothing is more amazing than plumbing. 04:18 Like we all get excited about plumbing. 04:20 So no. 04:21 - I know. 04:22 - Technology plumbing is cool. 04:25 - Yeah. 04:25 I mean, it's one of those things too, that look at how fast it grew. 04:28 Think about it like last year at this time, like at the time when we were recording 04:32 the work item episode, MCP did not exist. 04:34 - Yeah. 04:34 - Just not a thing. 04:35 - That's wild. 04:35 - And now everybody's talking about MCP. 04:38 Like you talk to any big companies, you talk to like the banks, the healthcare, the gaming, 04:44 Like everybody, everybody cares about MCP. 04:46 That's great. 04:46 It's very great. 04:48 We're going to dive into it. 04:49 Before we do, let's dive into you. 04:51 Give us a quick background on yourself. 04:53 Absolutely. 04:53 So I am Den Delamarski. 04:54 I am a principal product engineer at Microsoft. 04:57 I work in the core AI division. 05:00 So we're focusing on, as the name suggests, AI stuff, but applied to developers. 05:06 So I'm very, very heavily in the developer ecosystem. 05:09 And I'm one of the core maintainers of the Model Converse Protocol. 05:13 So I say one of because there's many of us. 05:15 It's not just me. 05:16 There's many wonderful, talented people way smarter than me. 05:19 And yeah, that's a short intro. 05:21 Okay. 05:22 So when we talk about MCP, you're one of the people helping build it. 05:26 That's incredible. 05:27 That is correct. 05:29 Yeah. 05:29 I try to contribute as much as I can. 05:31 Well, you know, before we move on, just how'd you get into that position? 05:36 Oh, it all started with one of the things that was actually near and dear to my heart, 05:40 which is security and authorization. 05:41 So when MCP first came out, it had a auth spec. 05:45 So we see on the screen right now, Michael is showing the kind of the model context vertical 05:49 specification page. 05:51 But when MCP first started, it had essentially a spec that outlines how to do authorization 05:57 for MCP servers. 05:58 And that spec was a good start, but it made a lot of assumptions about the infrastructure 06:04 and the tooling and how developers build MCP servers that were, I want to say, a little 06:08 flawed at scale. 06:09 So my thought was like, oh, I'll just get some smart people with me and we'll help rewrite this. 06:15 And we asked the MCP folks at Anthropic and they said yes. 06:18 And so we did. 06:19 And I basically like incorporate all the feedback and iterated on it. 06:23 And then again, it's a massive community effort. 06:26 We pushed it out and got it merged in the June version of the protocol. 06:29 And then the folks at Anthropic just reached out and said, hey, do you want to help shape the protocol? 06:35 And here I am helping shape the protocol. 06:37 You seem to know what you're talking about and you sure are participating a lot. Why don't you just hang around? 06:41 Yeah, basically. 06:42 Okay, that's great. And you work at Microsoft. What do you do there? 06:46 That is correct. At Microsoft, I work on developer tools. So think like if you ever use Copilot, if you ever use any, oh, by the way, GitHub spec kit for folks that have not heard about it, we released it like last month. 06:58 But that's something that I helped drive and help maintain is how do you do spec driven development with agentic tools, agentic coding tools? 07:06 Yeah, that's what I do. 07:07 Okay, cool. 07:08 So something that I've started to do a lot when I'm involving AI, I go in like spurts. 07:14 I'll work for a long time, just sort of writing regular. 07:16 And then I'm like, ah, this is really a lot of drudgery, not critical or central to what I'm doing. 07:21 Let me just uncork some agentic AI on it and let it go. 07:25 But one of the things I've started doing a lot, and it has to do with the spec thing that you've touched on here, 07:30 is I will force, I'll pick a really high level model, like a complex smart model. 07:36 And I'll say, I want to plan this out. 07:39 I've given you some ideas, look at the code and let's create a detailed plan 07:42 of what you're gonna do. 07:43 And I'll have it write a markdown file. 07:45 And even though a lot of my projects, I have just a plans folder 07:47 and it's just full of all these different projects. 07:49 You know, maybe they're sort of equivalent to a PR in the end. 07:52 - Yeah. 07:52 - And I'll plan that out really well. 07:54 Then I'll switch it down to a lower model, to a new context and say, let's just do phase one. 07:58 Let's do phase two and knock it out. 08:00 That sounds like a Michael just made up some stuff equivalent of the spec based programming. 08:06 Is that right? 08:06 Like, how does that compare to what you're talking about here? 08:09 It's close. 08:10 It's very close. 08:10 And this is where when when people talk about spec driven development, I want to emphasize 08:14 the fact that there's no one correct approach. 08:17 Like people think that it's like, oh, I'm just going to wait for whatever company is 08:20 going to come out and come up with the right thing. 08:21 Like it all depends on your experience. 08:23 It depends on your models. 08:25 The spec kit project that we launched is our hypothesis, our experiment on how we believe 08:30 And what it does is basically what do you describe? 08:32 You start with a spec. 08:34 You start outlining what and why I am building. 08:37 Then you focus on the technical implementation plan, which is like, OK, now what technology stack I'm using here. 08:43 And then you break that down into tasks, which are basically just consumable chunks that the AI can go and either iteratively or in parallel execute and build the stuff that you want to build. 08:54 So all of it, again, is still an experiment. 08:56 So I'm not by any stretch claiming that what we have is the end of it all or the right way to do this. 09:02 There's many, many ways to do this. 09:04 Okay. 09:04 And you even over on DevBlogs wrote, diving into spec-driven development with GitHub SpecKit. 09:11 That is correct. 09:11 There's also a GitHub blog that I highly recommend folks check out. 09:14 It's actually on the github.blog. 09:17 So you can go there and look for, there you go. 09:19 It's called Spec-Driven Development with AI. 09:21 Get started with a new open source toolkit. 09:23 And we do have an open source toolkit. 09:25 All right. So how is this different than just what I've done? I know I've seen this before. 09:29 Yeah. Okay. Yeah. It just, all it does is think of it as this is the recipe book, 09:34 right? Like if you decided to like, Oh, I want to cook up a new application and you're like, 09:39 well, what's the recipe? Like this bundles the recipe for instance. And by the way, 09:43 here's the box set of ingredients that you can just use to build this. That's what this is. 09:47 That's SpecKit. Okay. Well, very exciting. Let's maybe start to get into the main topic though. 09:54 So MCP servers. 09:56 I've heard this put out as sort of an analogy to the LSP, which I know is, I first heard of it in VS Code. 10:05 I don't know if it came from VS Code. 10:07 Maybe it did, but it's the thing that allows so many different editors to plug into tooling 10:14 like PyLance or Powerfly or ty or a bunch of cool things are coming out around here, 10:19 different implementations of LSPs. 10:22 And I've heard that MCPs are kind of like that for AI. 10:26 Maybe contrast those a bit for people. 10:30 Yeah. 10:30 I mean, if you look at the MCP specification, if you look through the website 10:34 and just peruse through the documentation, you might have like faint echoes of LSP design decisions, 10:40 faint echoes of kind of the LSP architecture. 10:42 But yes, basically think of it this way. 10:44 The way the folks at Anthropic have been describing it, it is USB-C for AI. 10:50 And when I say that is the problem with a lot of the LLMs, a lot of the modern models is the fact that it takes a some amount of time to train them, which means that inherently they get locked into a specific training date, if you will. 11:06 So the corpus of knowledge that gets embedded in them gets locked at a certain date. 11:11 And when you talk to a lot of enterprise customers, you talk to a lot of customers in the wild, 11:15 it doesn't need to be enterprise, by the way, it could be startups, could be hobbies, developers 11:18 like, well, I want to use AI with this fresh data that I have. 11:23 Maybe I have, I don't know, a Dropbox account and I want to use AI to sort my files. 11:27 Or maybe I want to use some data inside Salesforce to go and help me analyze my sales and find 11:34 out outliers and maybe customers I want to focus on. And I just interviewed the people from Nice Guy, 11:40 Nice GUI, and they build robots that cruise around in architectural areas. Like what maybe I want a, 11:48 I want some way to like ask AI, look at how the robots are doing now and then, or see if they're 11:55 busy, find a free one, right? That might be a thing, huh? Yeah. Yeah, no, for sure. Exactly. 12:00 It's like any kind of live data or managed data, something that is more dynamic than the corpus of knowledge that is embedded in these models by default. 12:09 And for those, if I would ask you like, OK, well, let's imagine a world where MCP does not exist. 12:14 How would you go about plugging this data in like into your LLM? Right. 12:19 And like there's different ways to do this. Like people have done like the rags. 12:25 People have done, you know, dump like CSV files and then be like, oh, analyze the CSV file and all these like hacky solutions. 12:31 But it feels like it's not universal. 12:34 It doesn't really work for all cases. 12:35 And something that you've done in one LLM doesn't work in another. 12:39 And now you're locked into this environment. 12:41 So it becomes very hard to manage. 12:43 So MCP is essentially the answer to that. 12:45 MCP says, look, we don't care what data you're connecting to, what applications, what actions. 12:50 we provide you a universal interface by which every single LLM, every single client that 12:57 understands MCP can invoke those primitives, get the data and embed the data in the context that 13:02 you're operating in. And that's another thing, important thing. People think of MCP as the data 13:06 connector, but it's not only a data connector. It's a, I want to call it like a primitive connector 13:11 because you can use MCP with a lot of wonderful things that folks have probably seen already. 13:15 Like I, my favorite example here is Blender MCP. Like for folks that don't know, Blender is a 3D 13:20 modeling tool. And there's an MCP server by which you can actually guide an LLM saying like, I am 13:25 building this like medieval scene with a dragon and the lighting and so, and it goes and it just, 13:30 it builds it for you, right? Through this MCP and MCP is the connective layer between Blender, 13:36 which has its own native API. And then there's the MCP server that the LLM knows how to talk to, 13:42 right? Because the LLM wouldn't know how to like, okay, how do you talk to Blender? How do you, 13:45 how do you go and set up the plugin and whatever the web sockets, whatever they might be using, 13:50 It's super complex, so it needs expertise, but an MCP server is essentially saying, I have these set of primitives that the LLM can invoke at any time, like create polygon or create scene or create sphere, and then based on that information, go and iterate on it. 14:03 So MCP is that adapter. 14:05 Yeah, I see. 14:06 So the LLM or agentic AI or whatever that you're working with, it says, all right, I'm going to talk to Blender. 14:12 Blender says, I have these core ideas, these core building blocks. 14:15 it sort of turns it more into Lego instead of just I'm going to have a saw or whatever I can 14:21 go. Exactly. Okay, I have spheres, I have cylinders, I have squares, I have shading. 14:28 They've asked me to do this. What can I build composing that sort of? Exactly. Precisely. Right. 14:33 So it's you're operating on a set of primitives, right? And this is where you don't even need to 14:37 expose the entirety of the surface of blender API's. You can just say like, oh, I want to have like, 14:42 there's the 10 primitives that I think are the most valuable. 14:44 I'm going to go ahead and use those. 14:46 And out of those, you compose things. 14:48 And maybe there's an advantage to that too, right? 14:50 Maybe you're like, I want to use Blender to create 2D scenes. 14:53 So I'm only going to expose stuff or rotations or whatever that preserves some sort of 2D view of the thing. 14:59 Like it's, we're doing CAD where it's top down from the side. 15:02 Like those are the ways you're going to look at. 15:03 You can't arbitrarily rotate it. 15:04 Yeah. 15:05 So yeah, so essentially like the MCP servers in this case act as a universal translation layer between whatever's downstream 15:11 of the MC server, which can be an application, an API, a database, like anything. And the client, 15:18 which knows like, I know how to talk to MCP and nothing else. I have no idea what's behind. I 15:22 don't know what the REST API you have, what's the authentication authorization logic, 15:26 just give an MCP server. Okay. It sounds a little bit like an API. And by API, I mean, 15:32 yes, most general sense of the word not, oh, it's a REST API. And it makes sure it uses the verbs 15:38 this way. I mean, like anything that you you could sort of call and either get data or cause an 15:43 action that could be a REST API, but it could just be, you know, an OS level API or some something 15:49 like that. Yeah, right. Yeah, totally. I mean, it's all it is just a connective layer. So yeah, 15:55 and people often ask like, well, couldn't you do this with like REST APIs? Couldn't you do this 16:00 with a GraphQL APIs instead? Right? Because like, it's been invented. Why are we creating new 16:05 things. But the thing about this is, even if you look in the world of REST APIs, like think about 16:10 the last time you worked with a REST API from some vendor and then switched another REST API from 16:15 someone, how much of that knowledge was like one-to-one reused or the infrastructure that 16:19 you built or authentication logic? You have like, you have these like 17 different dances by which 16:23 you get the token, right? And MCP essentially is the opinionated version of saying, no, this is how 16:31 you do auth. This is how you do message passing between entities. This is how you expose primitives. 16:37 It's a highly opinionated stack. This portion of Talk Python and me is brought to you by Sentry's 16:43 AI agent monitoring. Are you building AI capabilities into your Python applications? 16:49 Whether you're using open AI, local LLMs, or something else, visibility into your AI agent's 16:55 behavior, performance, and cost is critical. You will definitely want to give Sentry's brand new 17:01 AI agent monitoring a look. 17:03 AI agent monitoring gives you transparent observability into every step of your AI features 17:09 so you can debug, optimize, and control the cost with confidence. 17:14 You'll get full observability into every step of your AI agent. 17:17 That is model calls, prompts, external tool usage, and custom logic steps. 17:23 AI agent monitoring captures every step of an AI agent's workflow 17:27 from the user's input to the final response. 17:30 And your app will have a dedicated AI agent's dashboard showing traces and timelines for each agent run. 17:37 You'll get alerts on model errors, latency spikes, token usage surges, and API failures protecting both performance and cost. 17:46 It's plug-and-play Python SDK integration. 17:49 Open AI for now for Django, Flask, and FastAPI apps with more AI platforms coming soon. 17:55 In summary, AI agent monitoring turns the often black box behavior of AI in your app 18:01 into transparent, debuggable processes. 18:05 If you're adding AI capabilities to your Python app, give Sentry's AI agent monitoring the look. 18:10 Just visit talkpython.fm/sentry agents to get started and be sure to use our code, TALKPYTHON, one word, all caps. 18:19 The link is in your podcast player's show notes. 18:22 Thank you to Sentry for supporting Talk Python and me. 18:25 And so then once implemented them, we call them the hosts like VS Code or PyCharm or cursor or whatever, cloud code, it knows, all right, here's how I inspect the capabilities of this thing. 18:38 Here's how I stream back the responses if it's going to take it 10 minutes to do what I asked it. 18:43 This is how you do it with streaming HTTP APIs and so on. 18:46 Precisely, right? 18:47 Because you only need to then implement once. 18:50 And especially if you use one of the existing MCP SDKs that we're going to be talking down 18:54 the line, like that's the core value prop is like you do it once and it just works. 18:59 You don't need to worry about like, oh, but this other MCP server decided to implement 19:02 their auth in a completely different way. 19:04 What do I do now? 19:05 Yeah. 19:05 Now, if I want to build one of these things, does it does it have to be implemented in 19:11 an LLM or can I build just a traditional FastAPI API that ultimately does queries against 19:19 a database with no, no prompt? 19:22 Yeah, no, I mean, like MCP servers themselves are just essentially entities that they're 19:29 capable of exchanging JSON RPC messages. 19:31 Like you can absolutely, you can write a client that is completely detached from an LLM and 19:35 just invokes tools, right? 19:36 Okay. 19:37 Right. 19:37 Awesome. 19:38 You can, if you want to, I don't know why you would do that, but you 19:41 Absolutely can. 19:42 I'm sure people have a reason. 19:44 So I see a comment out in the audience from Frankie about RAG. 19:48 And also, you mentioned RAG at the beginning. 19:51 You say, well, maybe RAG is not working for you. 19:53 Let's just sort of contrast that a bit, right? 19:56 Like, maybe not everyone knows what RAG is. 19:58 Retrieval Augmented Generation. 20:00 What is this? 20:01 Yeah, essentially, if you have a way for you to optimize, basically, the context for the LLM. 20:07 I'll put it this way. 20:09 in very layman terms, it's like, I have a code base. 20:12 I have a code base that has a number of entities like classes and functions and everything. 20:19 And in a rag, you're essentially building a vector database 20:23 that says like, okay, here's the list of things that exist. 20:26 And then the LLM, you can go and query this thing and find out what exists in this code base. 20:31 So if you make decisions about like, I want to build a authorization component, 20:35 how do I do this? 20:36 Like, okay, well, it can build out that context for itself. 20:39 This is kind of the very basic idea behind the rag. 20:42 Got it. 20:43 So instead of trying to put all the information just into a prompt, 20:46 it has to read every time you can kind of additionally train it on these things 20:50 and then keep the question shorter because it knows the details. 20:54 Right. 20:54 You essentially have a knowledge base that's outside of the primary training set. 20:59 Like that's the core value prop of this is you're augmenting the LLM 21:03 with additional knowledge that you have in the context that you're operating in. 21:07 Okay. So something I've wanted to build for a while, and I do intend to, but we'll see if I 21:13 ever get there, is something where people could go and have like an AI conversation with this 21:18 episode, for example, right? With something on the podcast, I've got 10 years of transcripts. 21:23 Yeah. 21:23 You know, like over a million words, I'm pretty sure. That doesn't fit in most contexts. And 21:33 thing for Talk Python. Maybe there's an MCP angle that's really interesting. Like, what could I do 21:40 with MCPs in the podcast, do you think? MCPs in a podcast. So one of them, of course, is like querying 21:45 the data, which is I want to make sure that, you know, find me all the episodes where I ever talked 21:51 with Michael about AI, right? It could be one thing. I actually think that because of the richness of 21:59 the MCP capabilities, to me, when it comes to like podcasts, I envision a world where I can use MCP 22:03 piece to edit podcasts. That's my dream of this. And actually, like this is something that I've 22:08 been experimenting with because I haven't fully wrapped around kind of like how exactly that would 22:15 look like. But one of the things that I do, like as I'm sure you do when you edit the podcast, 22:18 you know, you have to go through it, generate the transcript, clean up things, then make sure that 22:24 you add timestamps, select the most interesting parts about the podcast. So can I potentially go 22:30 and say, okay, here's where my MP3 file is. 22:34 Can you go and generate transcript, clean it up, 22:37 and then find me the most interesting parts about this and then produce me a report 22:42 that I can then use to maybe like a HTML-based web app and I can just like a one-click save like publish, right? 22:49 And to me, like the value of maybe the MCP connector here is that maybe I can plug it in behind the scenes 22:54 with like FFmpeg to go and convert the MP3 into a WAV file and then use whisper to go and generate the transcript and then go and extract things for it 23:03 right and for a lot of these pieces of the tasks that you need to do you would imagine that you 23:07 would have a different tool inside my mc server which is tool is one of the primitives that 23:11 basically a lm invokes and it says oh let me generate the transcript and there's a tool that's 23:15 called generate transcript and it's gonna have that to produce a transcript and it's like okay 23:19 there's another tool that says yeah yes you could give just give the lm an episode number 200 or 23:26 something. And it could go to your podcast MCP server and say, transcript for 200, even if it 23:33 doesn't exist, it'll figure it out and generate it, that kind of stuff. Yes. And also the wonderful 23:38 thing about LLMs and MCP servers is that you're not actually using just one MCP server, right? So 23:43 I might have an MCP server for myself that is basically, like I said, the one that generates 23:47 transcripts, you know, creates a landing page in my podcast website. And then based on that content, 23:54 there's also next steps. Now I have an MP3, I want to upload that MP3 to Cloudflare, where I host my 23:59 podcast. So there may be a Cloudflare MCP server that the LM is going to invoke and say, I need to 24:04 now upload this. And then it's going to invoke the other MCP server, right? So you have this 24:08 basically stack of MCP servers that you can start using one with another. And that's where the 24:12 superpower comes from. Like you're not just using one application and saying like, okay, hold on, 24:15 let me, let me finish a task for podcast production. Then I'll do other things. Like it can chain 24:19 things together and then say, oh, and by the way, there's an MCP server maybe for audio conversion 24:23 that produces like 10 variations of the format. 24:26 Let me invoke that. 24:27 And then you're going through this process. 24:28 Yeah, I think that's one of the really big, hints at one of the really big differences 24:32 between just using a chat LLM versus some of the agent tool using types of things, right? 24:39 The ability to say, now I have to accomplish this task. 24:42 And I know I figured out there is some way I'm capable of accomplishing that, right? 24:46 Either that's to list a directory, to look for a file or to communicate with the Cloudflare MCP 24:52 that we talked about and so on. 24:55 Yeah, it's power is composability. 24:57 I'll put it this way. 24:58 It's the fact that you can compose things together and have them work together 25:01 based on the prompts that you have and scenarios that you have. 25:04 Okay, cool. 25:05 So imagining the Cloud Player MCP thing exists, your podcast preparation MCP thing exists. 25:12 How does my AI know? 25:14 Let's keep it real basic. 25:15 Let's say I'm using Claude Code, but we could plug this into others, 25:19 but just even something just terminal-based, no UI or whatever, like, yeah, it just is it going to discover them just out of the blue? 25:26 Probably not all of them. You got to point it at them. And yeah, how does it know which ones it's 25:30 allowed to use in this context? Right? Like, how do I get it so I can actually use one of these? 25:34 And we'll talk about maybe building them. Yeah. So for MCP servers themselves, you add them 25:39 explicitly to your host or your client, whatever that might be. VS Code, cloud code, cloud desktop, 25:44 doesn't matter. So you explicitly say, I want to use my podcast MCP. I want to be using my 25:49 Cloudflare MCP server. I want to use my, I don't know, Descript MCP servers to remove the ums and 25:54 uhs from the podcast, right? So you would essentially go through some means in that client, 25:59 on your client of choice, to go and add those MCP servers. Now, the question is, how do you 26:03 discover those MCP servers? So there's various places where you can go to. We just launched the 26:08 MCP registry that is nothing short other than an API that indexes all of the available MCP servers 26:15 that are out there, right? 26:17 So we're looking right now at a blog post on the MCP blog. 26:20 It's called Introducing the MCP Registry that got published September 8th of this year. 26:24 So not that long ago, but basically- 26:26 22 days or something like that. 26:28 And when you say we, you're talking the official 26:30 model context protocol.io working group. 26:33 Yes, yes. 26:34 The model context protocol folks, and there's a bunch of them 26:36 that were specifically focused on the registry, right? 26:39 And you see them in the authors like David Sariapara, Adam Jones, 26:44 But they essentially were in charge of kind of building this out. 26:47 And the registry is a centralized API, essentially, that aggregates an index of MCP servers that are out there. 26:54 So you can use the registry inside your client, whatever client you might be using, to find MCP servers for what you want. 27:00 Maybe there is a Playwright MCP server. 27:02 Maybe there is a Perplexity MCP server. 27:05 So it's all coming from the registry. 27:06 Okay. 27:07 Sounds a little bit like Docker Hub. 27:09 Kind of. 27:10 Yes. 27:10 And just like Docker Hub, you actually don't need Docker Hub to install an MCP server. 27:14 or in this case, like a Docker container, right? 27:15 Like you can just go to random GitHub repos and find somebody to build an MCV server 27:19 for what you're trying to do, and you can just plug it in. 27:21 - Yeah, interesting. 27:22 Yeah, that's how I use Docker Hub by not using Docker Hub for all the stuff I build, 27:26 but you know, I get the foundations. 27:27 - I know it exists. 27:29 - Exactly, I'm like, ah, but I'm gonna build it here. 27:32 It also has the concept of public and private registries. 27:35 - Yes, yeah. 27:36 So public registry is essentially something that like GitHub, 27:40 by the way, maintains their own registry, right? 27:42 So it's public and you can just go and discover MCP servers through the GitHub registry or the public registry. 27:48 Also, we know that MCP servers are used within different companies. 27:51 You might have, let's say, some data that you're locking in behind seven gates that only certain people can access. 27:59 You can build internal MCP servers. 28:01 And for those things, you ship internal private registries where you can say, no, no, no. 28:06 I want my folks in my company to only access these servers and nothing else. 28:11 Right. 28:11 Sure. 28:12 Yeah, that makes sense. Is there a place that I can go to the model context protocol registry, the MCP registry and like browse it like you can? 28:20 Yeah. Yeah. So right now you can't browse it through a UI, but you can look at other registries that can consume some of the content from here. 28:29 So like I believe GitHub registry is one of the consumers. 28:32 So you can look at I think it's gethub.com slash MCP. 28:35 There we go. 28:36 Yeah. 28:36 Okay. 28:37 And you can see some of the registries and you can see like if you click on one of the install buttons is going to like allow you 28:42 to take it directly into like VS Code and then just bring it in and install it 28:46 in the context of your editor. 28:47 - Okay, yeah, very nice. 28:49 So some of these are like web crawling, Notion. 28:52 Okay, I know Notion just added a big agentic AI thing and I've seen a lot of pushback. 28:58 There's probably a lot of happy users who just use it, but people are like, why is this in my way? 29:02 I just wanna work with this. 29:04 But you know, if you were, it'd be really cool to maybe plug that in instead of going, 29:08 we're gonna try to use the API to download this embedded database with the information. 29:13 - Exactly. 29:14 - Like you just talk to it, right? 29:15 - Exactly. 29:16 That's again, what I like about MCP is that if I want to connect to Notion to get my notebook 29:22 and some notes from my standup meetings, I don't need to worry about how they structure their API 29:26 and how to use auth or something. 29:27 Just install the Notion MCP and then ask the alum, pull the latest notes and summarize them for me. 29:32 And then it's gonna know. 29:33 - It's their LinkedIn one. 29:34 Their API is so bad. 29:35 - Oh. 29:37 - Oh. 29:37 - Oh, it makes me sad. 29:39 For any LinkedIn people watching this, we need to have a LinkedIn MCP server. 29:42 - Yes, I think so. 29:44 It might save me. 29:45 Okay, very interesting here. 29:47 I think people should come here and just kind of poke around. 29:50 You can see there's a lot of, a lot of interesting things that I think might 29:53 spark some ideas. 29:55 - Yeah. 29:55 - As you start to play with it, you know, like Postman. 29:58 So I guess one of the problem, well, not one of the problems, 30:00 one of the things you're gonna want to deal with is, a lot of these I see here, 30:04 LaunchDarkly, Postman, Atlassian, Notion, and so on. 30:08 You got to pass things like I am this person. 30:11 Therefore, I want to see my information, not other people's or only public. 30:16 I got to see private info, but mine. 30:18 There's a whole security side. 30:19 And I think that's kind of how you got pulled into it, right? 30:22 Yeah. 30:22 Oh, yeah. 30:23 Yeah. 30:23 So for these things. 30:25 Yeah. 30:25 We just put like an API key in GitHub and you just check that in and just use that when you're trying to. 30:30 Don't do that. 30:31 Don't put API keys in GitHub and check them in. 30:33 What can be done, so starting with the latest spec of MCP that, again, shipped in June, there is a formal way for services to do authorization. 30:42 So it's based on OAuth, OAuth 2.1. 30:45 I know that there's people listening that's like, oh, no, did you just say OAuth? 30:49 I have to learn OAuth now. 30:50 You don't. 30:51 Again, there's a lot of libraries that do this. 30:53 If you're an MCP server developer, it's solved for you. 30:55 If you're an MCP server consumer, you don't even need to think about it. 30:58 So when you connect an MCP server, as a consumer, you'll essentially have the ability to log in with your credentials. 31:04 So if an MCP server, for example, for like we saw Chromo and we're like MongoDB, that's on the screen here. 31:10 If I use the MongoDB server and I want to connect to a database, usually they provide you a way to either one. 31:16 You go into your MCP server config and you say, I will give you an API key if your server is using an API key. 31:22 Or if it's using OAuth, then you can just essentially snap to using OAuth the standard flow. 31:28 Your client is going to bootstrap the authentication flow. 31:31 You're going to go to the box, enter your credentials, log in. 31:33 The client is going to store the tokens, and then you access the server with your credentials as you getting access to your data, not something else. 31:41 One thing that looks really interesting, and there's an example of it right here with the Nux. 31:46 Never written a Nux app in my life, but here we have. 31:49 I have one that helps you understand your Vite Nux app. 31:53 One of the things that I think could be really interesting and probably MCPs could play a really important role is we have these huge foundation models, OpenAI and Cloud Opus and so on, that are generally knowledgeable about the whole world and are big, expensive to train. 32:09 But I can see a future where we get good enough to have a bunch of small models. 32:13 Like this is the Vue.js model. 32:17 If you need to know Vue.js, it's as good as anything, but it runs on your computer in a gig of RAM because it's just trained so specifically on Vue. 32:26 And I feel like maybe you could MCP your way together like, well, I'm using this tech stack. 32:31 So we're going to click together a bunch of things that don't provide data, but provide information about what your architecture or something like that. 32:39 What do you think? 32:39 Yeah, I mean, I think it can go both ways, right? 32:41 Like there's a specialized model. 32:42 And there's an argument for saying that the more general scenarios would always work best. 32:47 Like there's, I think there's always two camps of those folks that I talk to. 32:50 I personally think that I think for certain things, there is a tremendous amount of value 32:55 for hyper centralized or hyper local models. 32:59 I'll give an example, right? 33:00 Like I want to organize the photos on my machine. 33:05 Like maybe I have a lot of duplicates that, you know, because when you take photos of your modern cell phones, 33:09 like just click, click, click, click. 33:10 and then you have like 10 images of your dog and you're like, they're kind of the same, 33:13 but I want to pick the best one. 33:14 Like from a privacy standpoint, like I don't want to send that off to some server 33:18 remotely somewhere with my photos, which, you know, there's like family photos. 33:22 There's all sorts of like stuff that I do not want to send off to some remote server. 33:25 For those things, I want to use a local model. 33:27 And maybe there's an MCP server that allows me to basically like, 33:30 oh, I can find the photos and then crop them and like add some metadata or remove metadata 33:35 or whatever I want to do, right? 33:36 So for those things, I absolutely see the value in these like local models 33:40 where I can just say, I want it to be very good at this one specific task and that task only. And I 33:45 will never use this photo model for web app creation, but photos is going to be darn good. 33:49 And I think there's a lot of value for that. And if you augment it with MCP, I think it's 33:53 superpowers right there. Yeah, it does seem like it could be. It could be this little step would 33:58 benefit from a local model, but I don't want to constrain the entire problem solving to a local 34:03 model. Right. I think that's kind of the problem. Like I use LM Studio a lot and I've got, for 34:08 For example, I have the open AI 20 billion parameter open weights model that I actually 34:13 program against. 34:14 And it does all sorts of cool stuff for me, but I don't use it for my general work because 34:17 it's either too slow because it's on my Mac mini or I just want something that is better, 34:23 right? 34:23 Yeah. 34:24 And so if you're going to just start a, like I'm using this model to solve this problem, 34:28 that might not be the final outcome where we end up, right? 34:34 This portion of Talk Python To Me is brought to you by Nordstellar. 34:37 Nordstellar is a threat exposure management platform from the Nord security family, 34:41 the folks behind NordVPN that combines dark web intelligence, session hijacking prevention, 34:47 brand abuse detection, and external attack service management. 34:51 Keeping your team and your company secure is a daunting challenge. 34:55 That's why you need Nordstellar on your side. 34:57 It's a comprehensive set of services, monitoring, and alerts to limit your exposure to breaches 35:03 and attacks and act instantly if something does happen. 35:07 Here's how it works. 35:08 Nordstellar detects compromised employee and consumer credentials. 35:12 It detects stolen authentication cookies found in InfoStealer logs and dark web sources 35:18 and flags compromised devices, reducing MFA bypass ATOs without extra code in your app. 35:24 Nordstellar scans the dark web for cyber threats targeting your company. 35:28 It monitors forums, markets, ransomware blogs, and over 25,000 cybercrime telegram channels 35:34 with alerting and searchable context you can route to Slack or your IRR tool. 35:39 Nordstellar adds brand and domain protection. 35:42 It detects cyber squats and lookalikes via visual, content similarity, and search transparency logs, 35:49 plus broader brand abuse takedowns across the web, social, and app stores to cut the phishing risk for your users. 35:56 They don't just alert you about impersonation, they file and manage the removals. 36:00 Finally, Nordstellar is developer-friendly. 36:03 It's available as a platform and an API. 36:06 No agents to install. 36:08 If security is important to you and your organization, check out Nordstellar. 36:11 Visit talkpython.fm/nordstellar. 36:13 The link is in your podcast player's show notes and on the episode page. 36:17 Please use our link, talkpython.fm/nordstellar so that they know that you heard about their service from us. 36:23 And you know what time of year it is. 36:25 It's late fall. 36:26 That means Black Friday is in play as well. 36:29 So the folks at Nordstellar gave us a coupon, BlackFriday20, that's BlackFriday, all one word, all caps, 20, two zero, that grants you 20% off. 36:38 So if you're going to sign up for them soon, go ahead and use BlackFriday20 as a code and you 36:43 might as well save 20%. It's good until December 10th, 2025. Thank you to the whole Nord security 36:50 team for supporting Talk Python To Me. For sure. And especially because for a lot of the generalized 36:55 models, you're like, no matter how you look at this, you're not going to have the computer 36:58 resources anywhere near what like open anthropic has right so like in terms of speed and quality 37:04 what are you going to get you might get some like fine-tuned examples where some scenarios work very 37:09 very well but i think ultimately if we look at the general use case these generalizable models 37:14 are going to be ahead yeah i i definitely agree as well but i hadn't really considered how mcps 37:19 might allow you to use the really high-end models to compose specialized not quite as generally 37:24 smart but specialized versions of different things it could be yeah mcp can do anything mcp again is 37:31 just it's it's a pipe what you do with that pipe is up to you yeah well let's talk about how one 37:37 might build such pipes with uh with python so there's actually a model context protocol github 37:45 organization within there they have the python dash sdk the official python sdk for the mcp servers 37:51 and clients. So that's also interesting, the clients bit. So maybe we could kind of like, 37:56 there's a lot of concepts and things here, and I don't want to dive too much into code, 38:01 but maybe we could work our way through some of the concepts and some of the steps of building 38:06 such a thing. Yeah, totally. Well, I mean, it all starts from just getting the SDK, right? And this 38:11 is for like anybody that's using Python. You can just get it through pip or uv. I'm a big fan of 38:16 the folks at Astral. I think they're doing a fantastic job with uv and uvx. Like I use it for 38:20 or get up spec kit. 38:21 So, you know, uv add MCP, MCP CLI. 38:25 And there you go, you can be on your way. 38:27 It's as simple as that. 38:28 - Yeah, okay, that'll do it. 38:29 And then, yeah, you can specify like the CLI options or whatever kind you want. 38:34 - Yeah, yeah. 38:35 And also it's using fast MCP. 38:37 Are you familiar with fast MCP? 38:39 - No, I know some projects with fast in it, but not MCP. 38:42 - Yeah, so fast MCP is basically, think of it like FastAPI for MCP. 38:46 It's essentially like allowing you to compose MCP servers faster because it has a lot of the primitives baked in. 38:51 So things like authorization, which can be like kind of a pain point, but if you use 38:55 Fast MCP, it makes it a little easier. 38:57 And Fast MCP is a integral part of the Python SDK story for the actual like official Python 39:03 SDK. 39:03 Right. 39:04 The programming model looks like it would feel quite familiar to anyone who knows the 39:10 Flask API or beyond. 39:12 I think it's just, you know, a little sidebar. 39:14 I think it's really interesting how Flask is quite popular, but it's also spawned almost 39:20 every single web app after it has kind of borrowed its programming model. 39:25 So even if you're not exactly using Flask, if you're using Litestar or FastAPI or whatever, 39:30 you're still kind of doing that kind of programming. 39:32 And it's the same here, right? 39:34 You create an MCP as the app, you say @mcp.tool or @mcp.prompt and you put these onto functions 39:42 and they now become webized. 39:44 Yeah. Isn't that like, okay, like I am not like I write Python, but I'm not a Python expert. I'm 39:50 sorry, Brett Cannon, if you're watching this. But like, we'll take that part out, don't we? 39:55 As a stream's life. That's okay. So like these, like the do you call in Python, do you call them 40:02 decorators? Or is it like attributes like in C#, it's attributes. Yeah, in C#, it's attributes. 40:07 You do it with square brackets. In Python, it's decorators and you do it with the @ symbol. 40:11 Okay, so the decorators themselves. Look at the simplicity of this. Look at the screen right now of a sample where we're looking at the actual Python SDK repo. And one of the samples, you literally have a Python function, you have def add, and there is your arguments, you would pass you a function, like two integers. And then all you need to do to make that a tool that an LM can invoke is just add that at mcp.tool decorator. That's it. You're not going and crafting elaborate JSON RPC envelopes and converters and all these things. 40:41 like all the stuff is done for you add a decorator boom you have a tool that's it yeah it's simple 40:45 it's yeah it's really really simple to program and there's actually some fairly complicated 40:50 data exchange stuff going on like streaming partial results as they come in because 40:56 we're all used to two things ai requests taking a real long time but b that you see the little dots 41:03 thinking thinking and periodically like some stuff that's coming by to like oh yeah okay i see where 41:07 it's going. I don't know what it's going to come up with, but at least we could see it's working, 41:11 right? So to sort of keep that flow going, you've got the streaming style, right? 41:15 Exactly. And all of this is like, again, I'm looking at the sample. It's so, 41:20 the way I would describe it as a delightful developer experience. If I'm a developer, 41:24 I focus on writing the core functions. I don't have to worry about like, well, 41:27 how do I make this into a tool? Put a decorator on. That's how you make it a tool. 41:31 Yeah. Excellent. So I have this server and you mentioned that it's fast, 41:37 FastAPI or flask like how do I host it once once I call run or whatever I do on it yeah then what 41:44 I know I probably don't put it straight on the internet maybe I do I don't know so there's two 41:47 types of servers that you can have you can have local mcp servers and local mcp servers are 41:53 essentially just a local application think of it running like a console app or like your regular 41:57 python script and what it does it there might be referred to you might hear like they're called stdio 42:02 for standard input output. 42:03 And it's using basically native OS constructs to talk between processes, right? 42:08 The MCP client and the server. 42:10 So again, it's still JSON RPC, but JSON RPC over SDDIO pipes. 42:14 So the other one is streamable HTTP. 42:17 And streamable HTTP, it's again, MCP server that can be hosted somewhere in the cloud. 42:22 It can be hosted on your own home lab server if you want to, and you give it an IP address. 42:26 You can be hosted in AWS or Azure, GCP, doesn't really matter. 42:31 So for those servers, the JSON-RPC messages are basically done through the HTTP pipe with some set of HTTP conventions. 42:37 That's kind of where it is. 42:39 There's no constraint as to where you have to host it. 42:42 It's whoever supports running Python can host your MCP server. 42:47 Right, okay. 42:48 So I could put it behind Nginx or Caddy or whatever. 42:51 Like toss it into a container and put it somewhere. 42:54 Like it's totally fine. 42:55 Okay. 42:55 You know, you talked about all these sort of different, like, private, but online, but not quite online, you know, with, like, HomeLab and stuff. 43:02 I just want to give a shout out to Tailscale. 43:05 Like, have you? 43:05 Oh, yes. 43:06 Have you Tailscaled lately? 43:07 Oh, it is so good. 43:08 It is wonderful. 43:09 I love Tailscale. 43:10 It's my go-to thing. 43:12 And I'll tell you this. 43:12 Like, do you remember the days when you had to work? 43:14 This episode is not sponsored by Tailscale, for the record. 43:18 Should be. 43:19 Should be. 43:19 They can reach out. 43:20 Yeah. 43:21 Yeah. 43:21 Hey, Tailscale. 43:22 Yeah. 43:23 Yeah. 43:23 Michael Talks is awesome. 43:24 You should sponsor it. 43:25 But anyway. 43:25 So TailScale is great. 43:26 Like, remember the olden days when you had to like set up an open VPN 43:29 and be like, let me generate the keys. 43:31 Let me email myself the key so I can open it on the iPhone 43:34 and then add the key and then go through this process. 43:36 And it's just like, oh, man, such a pain. 43:39 Such a pain. 43:40 TailScale, just like flip the switch and you're in. 43:42 Yeah. 43:43 Magic. 43:43 Or DynDNS where you... 43:46 Oh, yeah. 43:46 Because you have to bind your IP address to their domain 43:50 and then you have to run this agent to constantly update it. 43:53 Oh, yes. 43:54 Yeah, the agent goes down change. Well, then there's also all the NAT firewall and your local machine on your local 43:59 network change. You're like, no, it doesn't work. Oh, it's my machine on my, we had a power outage 44:04 when the router rebooted, I got a new IP. It just, it was so bad. And so why is this sidebar worth 44:10 going into here, folks? Because this is what's called an overlay network. And so you can put it 44:16 up on your iPhone, you can put it on your laptop, you can put it on your desktop, you can put it on 44:19 your Linux server if you want. And it basically exposes all of those things over a network that's 44:25 like a VPN, but the rest of your behavior is just not VPN. 44:29 It's just normal, but it just brings those in in just the most incredible way. 44:33 So for example, I have a high-end Mac mini here that I use for the streaming that I'm talking to you on now. 44:39 It has tons of RAM and it has a pro chip and stuff. 44:41 So I just have my one LLM and my database servers running there. 44:46 And when I'm doing dev work, instead of every, you know, my laptop, 44:50 my other machine always running in replica, it all just goes here to this. 44:54 And even if I'm in a coffee shop or I'm out for work, right? 44:57 As long as TailScale is running, I do a database query or an LLM call through an API and it 45:02 just hits this thing. 45:03 Yep. 45:03 Just as if I was here. 45:04 And it's glorious. 45:06 And all that's for free, right? 45:06 There's paid versions, but you can do a lot. 45:08 Yeah. 45:09 You can do a lot for free. 45:10 In their free tier, it's amazing. 45:12 And it's all WireGuard. 45:13 It's all using the most modern secure standards. 45:17 I'll say to me, if you want to access things like, oh, your security camera is at home, 45:21 you do not trust cloud providers to have access to your security home cameras, put them in your 45:26 local network and use tail scale. And then you can go somewhere, flip the switch in your phone, 45:30 boom, you can see your cameras from remote without exposing them to the broader internet. It's 45:33 amazing. You don't open up any ports on your router, nothing like that. So why am I going on 45:39 such an excited diversion? One, it's just so awesome. And I just recently discovered it this 45:43 year. So it's a thing, but it's relevant. If you've got an MCP server and you want to keep it local, 45:49 even local from your server back to like your company or something potentially, 45:54 you could hide all that stuff behind tail scale. 45:57 It's like transparently available, but also there's, there's no ports. 46:01 There's no open internet. 46:02 The easiest way to secure stuff is to just not let the internet have at it. 46:06 Yeah. 46:06 Yep. 46:07 No, exactly. 46:07 This is what I've been actually doing with one of my friends who was setting up a 46:11 home lab and they were experimenting with some of the MCP servers for like, I believe 46:15 it was like setting up for like a Minecraft server. 46:17 And we just tossed them on the same server. 46:19 And because it's tail scale and then I connect them to the clients with a IP that tail scale gives me, 46:24 it just magically works. 46:25 And I didn't need to expose this to the internet. 46:27 I didn't need to pay for any cloud providers in somebody's home lab. 46:30 It's just there. 46:31 - Yeah, yeah. 46:31 And you don't need to use SSH across it. 46:33 Like you can just, it's just there. 46:35 It's all super, super good. 46:36 Okay, back, back to what I was asking. 46:39 - Back to MCP. 46:40 - Back to MCP, but I was asking, you know, how do you run it? 46:43 And you're like, I could, we could run it on our home lab or on a Raspberry Pi or something, right? 46:47 this tail scale thing is a way to sort of really nicely make that available to 46:51 you, make that available to your, your AI agents or whatever without going, well, 46:57 now how do I host it on like a server for real? Yeah. Okay. 47:01 So let's see. That is the registry. There we go. 47:04 So I want to talk about a couple of things. We talked about tools. Yeah. 47:08 And we talked about there's prompts, there's resources. 47:11 Let's maybe go through each one real quick. 47:13 These are all just decorators you put on functions, but they're all, 47:16 They're slightly different. 47:17 Yeah. 47:17 What is the purpose of a tool and why would I do that? 47:19 Yeah. 47:20 A tool basically is a function call, right? 47:22 It's like your tool equals function. 47:25 That's the way I describe it. 47:26 Like that's basically like, hey, I want the LLM to go do something. 47:29 What does it need to do? 47:30 And this is where like get weather, give me the sum. 47:34 It needs to go and do this. 47:35 This is what a tool is. 47:37 It's a primitive that does something. 47:39 Insert record into database or whatever. 47:40 This looks like you could probably find and replace Fast MCP with FastAPI and tool with 47:47 get. 47:47 Yeah. 47:47 Yeah. 47:48 And you or a post or something. 47:50 And you might be able to pretty much that is kind of the closest match, right? 47:54 Yeah, exactly. 47:55 Yep. 47:55 Yep. 47:55 That's that's basically it. 47:56 I want to invoke some kind of action. 47:59 Go do that action for me. 48:00 Right. 48:00 And at least in the examples, there's no AI in the action. 48:04 It's just. 48:05 No. 48:06 Just an AI. 48:06 The AI knows that it needs to invoke the action. 48:08 Like if I go to the LLM and say, send an email to Michael that says the podcast was awesome. 48:13 And then it's going to go in and say, oh, let me go find the tool that is capable of sending emails. 48:19 Oh, there's a tool from like, I don't know, like MailChimp. 48:21 Okay, let me go do that. 48:23 There's a tool in the MailChimp MCP server that says send email. 48:26 That sounds great. 48:27 I'm going to use that to send the email, right? 48:29 And that tool itself doesn't use AI behind the scenes. 48:31 It's just like, it's just going to do SMTP send email. 48:34 That's all it does. 48:35 Yeah. 48:35 Awesome. 48:36 It also has other examples of data exchange along the way, I guess. 48:41 Absolutely. 48:41 And you can pass in this context, and then the context can start pushing updates and information back. 48:49 Yes. 48:49 To the user, right? 48:50 And report progress back. 48:52 So, for example, if your email takes like seven hops, it's like, okay, let me first connect to the SMTP server. 48:57 Let me then verify the credentials. 48:58 Like, you can encode that basically if you implement that. 49:02 You might not, but you can implement progress reporting so that the client knows like, 49:06 oh, you're like 30% through your task or you're like 40% through your task now 49:10 because it reports on the progress of what you're doing. 49:13 - Yeah, super cool. 49:14 You can also do structured output, which is pretty interesting. 49:17 And there's many ways in which it can be done, but the number one way, 49:22 as in if it was a ordered list, the first thing would be Pydantic models, right? 49:27 Carrying on the FastAPI analogy here, right? 49:30 - Yep, yep. 49:31 For a lot of these things, again, it's very, like if you're a Python developer, 49:34 a lot of these concepts are gonna be very much familiar to you. 49:36 - Yeah, I think one of the challenges people have often is like structured data versus like I got an LLM answer 49:43 and it's a little different every time and they upgrade the model from 5.1 to 5.15 49:50 and now it does something totally different. 49:52 Like how do I code against this, right? 49:54 And so using structured data can be a big bonus, right? 49:58 - Yeah. 49:58 - Okay, super cool. 50:00 Let's see prompts. Now it's starting to sound AI like. 50:03 Yeah. So this is basically like the description says prompts are reusable templates that help 50:08 elements interact with your server effectively. If you have a server that does, I don't know, 50:13 cooking recipes, it might provide prompts for like, what is a like, what are the steps for a recipe 50:19 and with substitutions where needed. So it allows you to basically pre cook prompts that your server 50:24 might be using. Okay, they might be passing these internally to? Yes, yeah. So 50:30 return to the host AI, you know, there's a lot of AIs involved here. 50:34 Right, you know, essentially like, you're exposing prompt templates, like that's what it is, 50:38 like, and saying like, oh, if you're a user, if you're looking for like, creating a recipe, 50:42 this is a template for that prompt for a recipe. 50:44 Okay, cool. There's also a little bit of a UI component, which is interesting, 50:48 you can have a iconography representation of your actions. 50:53 Yeah, this is relatively new. But basically, for some of them, like, 50:56 bake in some of the icons to just make it easier to differentiate between different actions. Because 51:00 especially again, like different servers can have different tools and there are many tools. And how 51:04 do you like just parse the strings? Like just look at iconography. Yeah. Another thing that it has 51:09 built in support for is working with images. So that's pretty wild. I yeah, I noticed that for a 51:15 lot of the stuff, it's also like it's baked into the these are not necessarily like MCP spec constructs. 51:19 This is more like how the Python SDK exposes them and allows you to operate on them, right? 51:23 because like the fundamental constructs, the primitives are we have tools, 51:27 we have prompts, there is resources, which is another one. 51:30 And the resources is, allows the LM to basically think of it 51:34 as how do you refer to databases or files or entities within an API? 51:41 Those are, there's also elicitations as what Michael is showing right now on the screen. 51:45 So we have elicitations is a way for an MCP server to go to the client 51:49 and say, I want the client to provide me structured input on a specific question. 51:54 Like, hey, can you give us your date of birth? 51:58 And I expect a date. 52:00 Can you give me a date back exactly so I don't need to guess from the LLM context, right? 52:04 Or it can say like, you know, what kind of pet do you have? 52:07 And it can give you a list of options that you can actually have to pick from. 52:11 It's like, oh, dog, you know, pet, reptile, like dog, cat, reptile, whatever. 52:15 It allows you to have that structured controlled input that it's not just you're typing into the chat box, 52:21 but you're selecting from a list that the server asks you to. 52:25 So that's another neat thing that recently got added. 52:27 Yeah, that looks quite interesting. 52:29 And it has to do a little bit with the WebSocket type of exchange as well, right? 52:34 Not exactly, but it's going along. 52:37 You've asked it something. 52:38 While it's working on that, it's come back and it's asking you to give it more information to carry on. 52:42 Yes, exactly. 52:43 In that sense, right? 52:45 Yep, yep. 52:45 So could this be, I've worked on your request. 52:49 I've used the database MCP or whatever, and I've learned that there's 20 records. 52:54 Do you want to delete them like you asked or do you not want to delete them? 52:57 Yes, yes, exactly that. 52:58 Or it can say, hey, I found like 10 conflicting records. 53:02 Which ones do I need to delete? 53:03 And then you can help and basically do, yeah, right? 53:05 So it asks for structured input so that you don't have to have it guess from whatever you type in the chat. 53:11 Because if you type in the chat, it's like it's non-deterministic, right? 53:14 It could say, oh, delete all the records with the name John Doe. 53:18 And then it's like, oh, I'll delete everything with dough. 53:20 Because somehow, like, that sort of decides, like, oh, no, no, no. 53:23 Jane, come back. 53:24 Yeah. 53:26 So it adds a little bit more structure. 53:28 Yeah, got it. 53:29 And the programming model is super smooth here. 53:32 They did a great job. 53:33 So, for example, you might be doing this elicitation within a tool call. 53:39 And that's an async function, async web function. 53:41 And the way you do it is just await context that elicit some message and schema. 53:46 And then when the person responds, the async thing resumes and off you go, right? 53:51 There's not some nested callbacks and all that kind of business. 53:54 That's a very smooth developer experience. 53:56 I love it. 53:57 Yeah, it definitely is. 53:58 Okay, I do want to talk about some of the popular ones out there 54:04 through an awesome list because I'm just a sucker for awesome, awesome list. 54:08 But is there anything else that I feel that you feel like we should be covering here 54:12 on the SDK? 54:14 Yeah, there's a lot of great work done Python SDK and the FastMCP folks, 54:18 I would say like go through the repo. 54:21 It's getup.com slash model context protocol slash Python dash SDK. 54:26 Go there. 54:27 There's some great samples to get you started. 54:29 And again, we're always open to feedback. 54:31 So if something's like, oh, this was too confusing. 54:33 I didn't understand. 54:34 The team is very receptive to feedback. 54:36 So please let them know. 54:37 Yeah. 54:38 143 contributors. 54:40 Last release five days ago. 54:42 Bunch of PRs, right? 54:44 It looks like it's pretty open. 54:45 You know, Yeah. Oh, yeah. 54:46 Close PRs pretty open to people working. 54:49 Also, it looks kind of very beginner friendly in the sense that the issues are 54:55 tagged with lots of lots of stuff that you could search for, like needs motivation. 55:00 You know, you could go through and come up with some examples and help, even if 55:03 you're not an expert in the SDK, for example. 55:05 Absolutely. 55:06 And there's also, I believe the Python might be using the good first issue too. 55:10 So if you're, if you're a new contributor, you've never looked at it. 55:12 It's like, I like, don't be intimidated. 55:14 There's plenty of- 55:15 Good first issue. 55:16 Good first issue. 55:17 Like there's plenty of things that you can just drop in and see like, oh, I can help with that. 55:21 Yeah, love it. 55:22 Okay. 55:23 You too can be an AI developer. 55:24 I love it. 55:24 Now let's talk about awesome MCP servers. 55:27 Awesome MCP servers. 55:29 This comes to us from the very well-known PunkPi. 55:33 The person behind Glamour.ai. 55:36 Yeah, awesome. 55:37 And 72,000 GitHub stars, no joke. 55:41 So there may be a fad, but maybe people will stick around. 55:43 So this actually has support for a lot of different languages 55:47 and it's got scopes like is this cloud or local or embedded 55:50 and so on. 55:51 But then you scroll down. 55:53 Look at the list. 55:54 Massive. 55:55 The list is, I mean, look at the scroll bar. 55:58 It is massive. 55:59 Yeah, we keep scrolling and scrolling. 56:01 I don't know. 56:02 If I page down full speed and just pin page down, the pinch down button, it's something along the lines of like 56:09 five seconds just to get through the list. 56:11 And these are one per line. 56:12 Mm-hmm. 56:13 You know, it starts out as one should when they're building awesome lists with categories, right? 56:19 Command line, cloud platforms, biology medicine, and bioinformatics. 56:26 There's one for everything. 56:27 I know. 56:28 You want to just jump around a bit and we can see what's here when we riff on it? 56:31 Gaming. 56:32 MCP server for Unity 3D game engine integration for game to own. 56:36 That's kind of cool. 56:37 Go. 56:38 Unity MCP. 56:38 MCP chess. 56:40 An MCP server playing chess against LLMs. 56:42 Do you ever think of like, can I beat an LLM at chess? 56:46 And you want to like just get an MCP server to do that? 56:48 There is one for that. 56:49 I'm starting to feel like it's better to do the local models 56:51 for the chess playing against the one. 56:54 I don't want the really smart ones. 56:56 There's also chess MCP, which is, this is interesting. 56:59 It's not the same as the other one. 57:01 This is access your chess.com player data and records and other public info. 57:06 Yep. 57:06 Right. 57:07 That's kind of cool. 57:08 So if you wanted to say, hey, I'm building something and I would like access to sort of the Kaggle of chess players type of thing, right? 57:15 Like the list of competitive chess results. 57:18 Yeah. 57:18 That's kind of cool. 57:19 Yeah. 57:19 Yeah. 57:20 Yeah. 57:20 I personally have built one for Halo. 57:23 I'm a big fan of Halo, the video game. 57:24 Oh, yeah. 57:25 It's not on the list, which now I need to go and contribute to that list. 57:28 Let's do a PR. 57:30 Like, that's the thing that I have is basically analyze my Halo stats. 57:34 And I'll tell you what, the LLMs are getting really good at analyzing the stats. 57:38 You give them the data, they can make some conclusions. 57:40 Yeah, I bet. Let's just keep it really crazy. Let's do, I was going to do delivery. We'll do that in a moment. Marketing. 57:46 Marketing. Yeah. 57:48 Yeah. So I guess one of the things that looks, I'm after just a very quick first impression, like you're running ads on someone's platform or you're doing marketing on someone's platform, but you want visibility into how that's going. 58:00 So we've got the Facebook ads and PC server. 58:03 We've got the Google ads, MCP server, Amazon ads and so on. 58:07 Right. 58:08 But what else is, yeah, that sounds about like most of it there, I suppose. 58:11 But think of it this way. 58:12 Like if you connect several of these MCP servers to your client and then you connect them 58:17 to all your ads accounts and then say, how are my ads performing 58:20 and which ones of them are the best this past week? 58:23 Right. 58:23 Like I don't need to click around dashboards and figure out like the filters and everything. 58:26 Just ask the LLM, pull the data, make a conclusion. 58:29 Now, you still need to verify the conclusion that make sure it's not hallucinating things. 58:32 But nonetheless, it's kind of cool. 58:34 Yeah, it's very cool. 58:36 So one thing I know I realized now that we skipped over the Python SDK is we talked all 58:40 about the server. 58:41 What about client things? 58:42 If I wanted to create an MCP server that is effectively the composition of some other 58:48 MCP servers, could I do that? 58:49 You absolutely can. 58:51 Nothing stops you. 58:51 Like an MCP server can also act as an MCP client and then connect to other MCP servers. 58:57 Like there's no restriction to that, right? 58:59 Like it's basically, it's very composable. 59:01 And a client for all intents and purposes is basically an entity that can connect to an MCP server, 59:07 which can also be an MCP server. 59:08 It's kind of circular. 59:09 Yeah, yeah. 59:10 It's turtles all the way down, but MCP this time. 59:12 Yeah, it's AI turtles this time. 59:14 So delivery, we just have the DoorDash delivery MCP server. 59:17 Oh man, like who? 59:19 Claude, why is my food not here? 59:22 Have you ever seen those fail videos or whatever? 59:25 I watch weird YouTube stuff with my daughter sometimes and you'll see like cops delivering DoorDash. 59:31 I'm gonna say, sorry, we had to arrest your DoorDash delivery, 59:33 but we were pretty close. 59:34 So we thought we'd just go and deliver your food anyway. 59:36 I mean, I don't know what the server is gonna say, but it could say anything, you know? 59:40 - The police are on their way. 59:42 - Yeah. 59:42 People are generally really appreciative. 59:44 Like, well, thanks for getting me my dinner anyway. 59:47 Let's see what else is out here. 59:49 Got text to speech, which is interesting. 59:52 - Sports. 59:52 - Sports, hell yeah. 59:54 Oh, look at this, Strava. 59:55 Like if you're running or biking, you can use this also to analyze your data. 59:59 There's a lot of MCB servers for data analysis, which is kind of cool. 01:00:02 Okay, I don't even, this one, this is the one that appeals to me. 01:00:05 So Multivewer, this is actually not a thing that I would want, but I think it's interesting. 01:00:11 So Multivewer is a motorsports desktop client. 01:00:14 And what I think it does, it does for IndyCar, WAC, Formula One, 01:00:18 and even like the feeder classes. 01:00:20 I think what it lets you do is put up both an overlay of telemetry onto watching the live stream, 01:00:27 but also put the multiple people up in live streams at the same time or 01:00:32 something like that. 01:00:32 Right. 01:00:33 That's kind of cool. 01:00:34 That's cool. 01:00:35 So the, the reason I don't really like that is I don't watch any of those sports 01:00:38 live. 01:00:39 I record them. 01:00:39 And so I can then pause it and then skip the commercials. 01:00:42 And so this is like for a live stream sort of deal, but the MCP server, 01:00:46 it controls multi viewer for that. 01:00:49 So maybe you could set up an AI that is watching what's going on and switches the views around in the multi-viewer for you. 01:00:58 That's wild. 01:00:58 Or swaps to the most interesting telemetry at the specific moment. 01:01:02 Yeah, listen to the radio. 01:01:03 They start getting all frantic. 01:01:05 Like, all right, we're switching to that view. 01:01:08 Yeah, there's an NCC server for everything. 01:01:10 Like, this list is massive. 01:01:12 I'm actually like, every time I discover these things, like, we're looking at this right now, I was like, oh, I didn't know there was one for multi-viewer. 01:01:17 Like I didn't know what multiviewer is until we talked right now. 01:01:20 Yeah, but wouldn't that be a cool demo? 01:01:22 Yeah. 01:01:22 You know, at a conference, you're like, I know you've all seen the tic-tac-toe one, 01:01:27 but let me show you the final of F1. 01:01:30 Yeah, yeah, yeah. 01:01:30 Or something, right? 01:01:32 Very astute observation, because again, like there's a lot of these like hello world kind 01:01:35 of things like, oh, look, it's kind of neat. 01:01:36 It responded with a thing like, give me a real thing. 01:01:38 This is that real thing. 01:01:39 Yeah, yeah, that's, that's super neat. 01:01:41 All right. 01:01:42 I guess we've got the support one that Lassie and Jira quick chat. 01:01:46 It's whatever you want, right? 01:01:47 That's the one to reduce your boring work. 01:01:49 The GRI MCP server. 01:01:50 Like, you don't want to triage your bugs. 01:01:52 Just let the LLM do it for you. 01:01:53 Hey, can you go and find the things that are most important for me to work on today? 01:01:56 Give me the bug numbers. 01:01:58 Yeah. 01:01:58 Or if you see somebody assign a bug to me, close it. 01:02:01 Yeah. 01:02:01 Yeah, exactly. 01:02:02 Query all the bugs assigned to me, reassign them to somebody else. 01:02:09 Yeah, crazy. 01:02:10 Exactly. 01:02:11 Not a good fit for this person. 01:02:13 Yeah. 01:02:13 No, exactly. 01:02:14 These are the life hacks you learned only from this podcast. 01:02:17 That's right. 01:02:18 It's like, if it involves MCP servers and cool stuff I can code, give it to me. 01:02:22 Otherwise, send it somewhere else. 01:02:24 Send it somewhere else. 01:02:26 All right, Dan. 01:02:27 I think we're getting pretty close on time here in terms of what we got time to cover. 01:02:31 But this is super fun. 01:02:33 Maybe close things out for folks. 01:02:35 They want to get started with MCP servers, either building them, consuming them, building 01:02:39 and consuming them, plugging them into their tool chain. 01:02:43 What do you tell them? 01:02:43 Yeah. 01:02:44 So for folks that wanna build modelcontextprotocol.io, as simple as it gets, go there. 01:02:48 It has guides, tutorials, SDK starters, everything is there. 01:02:52 If you are a consumer of the MCPs and you wanna, hey, I wanna do this like awesome thing with MC servers. 01:02:58 First of all, the GitHub MCP registry that we showed earlier is one of those things 01:03:03 is github.com/mcp, go explore. 01:03:06 And then of course on GitHub, there's plenty of servers that are tagged with MCP. 01:03:10 You can also take a look there. 01:03:11 And there's other registries that also index MCP servers of all sorts, like Glama AI from 01:03:17 Punk Pie that we talked about before. 01:03:19 There's one such registry that you can also look at and see if there's anything that's 01:03:22 of interest. 01:03:24 I will say that as you are exploring MCP servers, exercise caution, just like you would exercise 01:03:30 with any other software and APIs and websites where you log in because the responsibility 01:03:36 is kind of on you to figure out what's safe, what's not. 01:03:40 If you have an MCP server that's like, oh, it's going to read all my iMessages and sort them by importance. 01:03:45 I'm like, yes. 01:03:47 And do you know who built that and where your messages are going? 01:03:50 So be careful. 01:03:51 Are they also scanning for credit card numbers? 01:03:53 Exactly. 01:03:54 Why not? 01:03:55 You messaged somebody with your social security number the other day. 01:03:57 Nice. 01:03:59 Yeah. 01:03:59 So be careful with those. 01:04:00 But I'd say, like, explore them. 01:04:01 And then we are working on formalizing discovery a bit better. 01:04:06 your clients like VS Code and Cursor and Cloud Desktop are going to become better and better 01:04:11 with more discoverability affordances. Awesome. All right. Thank you so much for coming on the 01:04:14 show. I learned a ton. I'm sure listeners did as well. And it was a lot of fun. Thank you for 01:04:19 having me. Yeah. See you later. Bye. This has been another episode of Talk Python To Me. 01:04:24 Thank you to our sponsors. Be sure to check out what they're offering. It really helps support 01:04:28 the show. This episode is sponsored by Posit Connect from the makers of Shiny. Publish, 01:04:34 share and deploy all of your data projects that you're creating using Python. Streamlit, Dash, 01:04:40 Shiny, Bokeh, FastAPI, Flask, Quarto, Reports, Dashboards, and APIs. Posit Connect supports all 01:04:47 of them. Try Posit Connect for free by going to talkpython.fm/Posit, P-O-S-I-T. 01:04:54 And it's brought to you by Nordstellar. Nordstellar is a threat exposure management platform 01:04:59 from the Nord security family, the folks behind NordVPN that combines dark web intelligence, 01:05:05 session hijacking prevention, brand and domain abuse detection, and external attack surface 01:05:11 management. Learn more and get started keeping your team safe at talkpython.fm/nordstellar. 01:05:18 If you or your team needs to learn Python, we have over 270 hours of beginner and advanced courses 01:05:24 on topics ranging from complete beginners to async code, Flask, Django, HTML, and even LLMs. 01:05:31 best of all there's not a subscription in sight browse the catalog at talkpython.fm 01:05:36 be sure to subscribe to the show open your favorite podcast player app search for python we should be 01:05:41 right at the top if you enjoy the geeky rap theme song you can download the full track the link is 01:05:46 your podcast player show notes this is your host michael kennedy thank you so much for listening i 01:05:51 really appreciate it now get out there and write some python code 01:06:06 I'm out. Copyright © PDX Web Properties, LLC 2015-2026. All Rights Reserved Made with in Portland, OR, USA

Certains moteurs de recherche ignorent purement et simplement le JavaScript mal optimisé, alors qu’un simple ajustement HTML peut changer le classement d’une page du tout au tout. Pourtant, des frameworks comme React dominent désormais de nombreux sites à fort trafic, malgré des défis persistants en matière d’indexation.Selon les dernières recommandations de Google, la structure du code source prime souvent sur la seule qualité du contenu. La compatibilité entre langages de programmation et robots d’indexation s’invite ainsi au cœur des stratégies SEO, avec des conséquences directes sur la visibilité organique. Le choix du langage de codage pèse lourd dans le dialogue entre vos pages web et les moteurs de recherche. Google, Bing ou Qwant accordent la priorité à la lisibilité du contenu et à la rapidité d’accès à l’information. Sur ce terrain, le HTML s’impose : il structure le site, trace des repères clairs, guide sans complexité les robots qui sillonnent le web. A découvrir également : Largeur idéale pour votre site web : comment la choisir ? JavaScript met l’accent sur l’expérience utilisateur et l’interactivité. Mais dès que trop de fonctionnalités lui sont confiées côté client, les robots d’indexation peuvent perdre leur chemin. Si le contenu apparaît trop tard ou se trouve masqué, la visibilité chute. Grandir sur le web grâce à une portion de Python côté serveur séduit de plus en plus pour générer du contenu pertinent, mais tout passe par une restitution HTML impeccable. À chaque technologie, son domaine de prédilection : A lire en complément : Zectayaznindus, miroir du web moderne : ce que ce mot révèle sur Google Savoir manier ces langages affine la présentation de l’information, favorise l’accessibilité et ouvre la porte à une indexation rapide. Pour viser haut, impossible d’ignorer les standards imposés par les moteurs, sans jamais oublier la fluidité de navigation. Atteindre les meilleures positions sur Google demande une combinaison pointue de technologies. Le socle reste le HTML : bien structuré, pensé pour guider robots et utilisateurs. Sans lui, les contenus même les plus inspirés passent à côté de leur public. JavaScript, quand il orchestre les animations et personnalise les applications web, devient un atout décisif. Mais il impose sa rigueur : si le contenu s’affiche trop tard, faute d’un bon rendu côté serveur, l’indexation s’en retrouve freinée. Miser sur le server-side rendering aide à contourner ces obstacles, offrant au robot tout ce qu’il est venu chercher. Python a pris sa place dans l’arsenal SEO : génération dynamique de pages, analyses en profondeur, gestion automatisée de la data. Il se glisse derrière chaque stratégie de contenu adaptée au marketing digital, tout comme PHP ou Java qui, moins médiatisés, gardent leurs fonctions clés dans les architectures solides. Pour y voir plus clair, voici une synthèse des forces de chaque langage : À chaque type de projet, s’assurer que le langage retenu communique efficacement avec les moteurs de recherche, qu’il hiérarchise clairement l’info et repousse les freins à l’accès instantané au contenu. La structure du code oriente directement la manière dont les moteurs classent et comprennent un site. Savoir dompter les balises HTML reste le socle d’un référencement solide. Balises meta, titres structurés, attributs alt judicieusement choisis : chaque détail compte quand il s’agit d’expliquer à un robot, ou à une personne en situation de handicap, ce que propose chaque page. L’enchaînement des balises, bien pensé, lisible et organisé, n’a rien d’accessoire. Un texte clarifié par son codage gagne en indexabilité. Sur de nombreux sites, cette exigence de clarté fait la différence. Il suffit parfois d’une structure soignée pour franchir une marche dans les résultats de recherche. Voici quelques principes techniques à intégrer pour renforcer la présence sur Google : Certains CMS proposent des automatismes sur la structuration, mais la main de l’humain demeure précieuse pour ajuster la finesse des balises. Un contenu repensé pour tous les utilisateurs, peaufiné côté technique, trace sa route vers de meilleures positions. La marche rapide du référencement naturel s’accélère, portée par l’irruption de l’intelligence artificielle et la prise en compte des comportements réels. En 2025, chaque adaptation à de nouveaux algorithmes peut signifier plusieurs places gagnées. Les robots des moteurs de recherche privilégient les sites qui assurent une expérience mobile irréprochable, une organisation des contenus limpide et des réponses précises aux besoins utilisateurs. Les mutations s’opèrent autour de trois axes principaux : Collecter, analyser, affiner : aujourd’hui, Python ou JavaScript associés à une analyse fine de l’audience donnent le rythme. Savoir anticiper les attentes, garantir une rapidité d’affichage et rester fidèle à l’intention de recherche forment le vrai terrain de la visibilité future. Ceux qui garderont la main sur la technique et l’agilité sur les usages se donneront toujours une longueur d’avance. Recherche Articles en vogue © 2025 | tic-et-net.org Sign in to your account Identifiant ou adresse e-mail Mot de passe Se souvenir de moi