AI That Acts: Odyssey-3, Helix 2.5, Jev & DeepMind’s New Institute

Odyssey-3 connects a world model to robot control. Figure takes Helix 2.5 into thirty unfamiliar homes. Jev brings fast decisions into agent workflows, while an experiment with communities of agents shows how a malicious message can influence behavior almost two days later. In this episode of Attention Span, we connect the week’s developments in AI’s ability to understand and act in the world. We explore Reka’s world-model plans, why handing a box to a person is brutally hard, and how FAMOS infers an object’s moving parts from incomplete observations. I also share my Dreamforce conversation with Salesforce’s AI research team about learning from feedback and updating models after deployment. We finish with Google’s household assistant CC, Edison Scientific and FutureHouse’s Millennium Problems for Biology, and the questions raised by the new DeepMind Institute. 👇 Which of these developments deserves a deep dive? Tell us in the comments. 🔳 SPONSORSHIP Partner with Turing Post to reach AI engineers, researchers, and builders: ks@turingpost.com 🔷 WORLD MODELS Odyssey-3 announcement and demonstrations https://odyssey.systems/introducing-odyssey-3 Reka’s omni-world-model research direction https://reka.ai/news/evolution-of-llms-omni-world-models 🔶 ROBOTS, GENERALIZATION & MEMORY Figure Helix 2.5: zero-shot tests in 30 homes https://www.figure.ai/news/helix-2-5-zero-shot-30-home-generalization Index human-behavior dataset, background https://www.figure.ai/news/introducing-index Agility Digit 5 https://www.agilityrobotics.com/solutions/digit-5 Agility’s safety interview https://www.youtube.com/watch?v=2dbyg67EtUA FAMOS paper https://arxiv.org/abs/2609.20817 FAMOS demonstrations https://kevinqu7.github.io/famos/ Workspace Models paper and task examples https://arxiv.org/html/2609.20820v1 🔹 DREAMFORCE & LEARNING FROM FEEDBACK Salesforce Koa announcement https://www.salesforce.com/news/press-releases/2026/09/15/koa-reasoning-model/ How Salesforce trained Koa https://www.salesforce.com/news/stories/why-we-post-trained-our-own-reasoning-model/ 🔸 EMERGENCE WORLD & AGENT MEMORY Season 2 research paper https://world.emergence.ai/publication/emergenceworld-s2.pdf Season 2 video https://www.youtube.com/watch?v=LTtbTEufPGA 🔹 JEV & FAST AGENT DECISIONS TypeSafe AI’s Jev announcement https://typesafe.ai/blog/introducing-system-one-models-and-jev Our detailed guide: Jev, RLCD, and the AI classifier https://www.turingpost.com/p/what-is-jev-rlcd Jev on Vercel AI Gateway https://vercel.com/changelog/typesafe-ai-jev-now-available-on-ai-gateway LangChain: building a harness with Jev https://www.langchain.com/blog/building-a-harness-with-jev 🔸 GOOGLE CC Google’s experimental agent for families and households https://blog.google/innovation-and-ai/models-and-research/google-labs/cc-expanding-to-groups/ 🔺 DEEPMIND INSTITUTE & SCIENTIFIC AMBITIONS Introducing the DeepMind Institute https://institute.deepmind.com/essays/introducing-the-deepmind-institute/ DMI essays https://institute.deepmind.com/#essays The Millennium Problems for Biology https://millenniumproblems.bio/ Organizations behind the Millennium Problems for Biology: Edison Scientific https://edisonscientific.com/ and FutureHouse https://www.futurehouse.org/ 🔻 MORE FROM TURING POST Newsletter, research coverage, and AI Builds AI https://www.turingpost.com/ Instagram: https://www.instagram.com/turingpost_tv TikTok: https://www.tiktok.com/@turingpost_tv Subscribe for Monday news digests and Attention Span deep dives into how AI works.