One Brain, Any Body
The opening shot: a single foundation model that can drive any robot body. Why "omni-bodied" intelligence, not one robot at a time, is the unlock, and what it means for everyone building in physical AI.
Thursday, July 16, 2026 · The Midway, San Francisco
Two stages, one day. Foundation models for robots, fleet economics, driverless freight, and the people actually shipping it.
Doors open. Grab coffee and breakfast, pick up your badge, and start the morning among the founders, engineers, and investors building the future of physical AI.
A quick welcome and the lay of the land before a full day of robotics and physical AI.
The opening shot: a single foundation model that can drive any robot body. Why "omni-bodied" intelligence, not one robot at a time, is the unlock, and what it means for everyone building in physical AI.
A fresh, billion-dollar counter-thesis: robots that learn to act by predicting what happens next, frame by frame, instead of being told in language. Teaching machines to anticipate the physical world.
The contrarian path to self-driving trucks: build the autonomy inside the truck maker, not as an outsider. The capital-disciplined road to driver-out freight.
From perceiving humans in 3D to self-driving to general robotics, why "world models" that imagine the physical world may be the connective tissue of all embodied AI.
Can AI actually rebuild American manufacturing, or is it still a demo? Three builders on what's shipping on the production line today, and what reshoring really takes.
Refuel and connect. Coffee, conversation, and a chance to find the people you came to meet.
The team teaching heavy machines to drive themselves. Why the construction site may be the next great frontier for real-world autonomy.
A rare founder who has put physical robots into messy real-world venues at fleet scale, and the hard-won economics of robots-as-a-service: what actually pencils, and what breaks.
Robots have to practice in simulated worlds before the real one, but building those worlds takes weeks. A look at generating endless, varied training environments from a simple prompt.
Everyone is racing to build robot brains, but the real bottleneck is the data, models, and compute that train them. Sharply different views on where it comes from, who owns the moat, and whether power and GPUs are the real cap on how big these models get.
A robotics pioneer who imagined "cloud robotics" 15 years ago now runs physical AI at warehouse scale, where embodied AI grows up, coordinating thousands of robots on the floor.
Harry Mellsop also steers the conversation.
Autonomy is not just cars. The robots quietly solving the last mile, flying and rolling, and the hardest validation problems in the field.
A lunch built for connection. Eat, network, and trade notes with the room.
The post-lunch jolt: live robots and hands on stage, and fast pitches from the startups building the next physical-AI products.
Navigation feels solved; grabbing a random box from a messy trailer still is not. The multi-billion-dollar frontier of manipulation and fulfillment, from the companies cracking it.
Past the highlight reel: laser-weeding farms, solar-building robots, layout robots on live construction slabs, and the safety layer that lets any fleet run. What "at scale" actually means.
The least-glamorous, most-decisive topic for founders: how you actually build and scale a team that ships hardware, across retail, seafood, and the factory floor.
The investors writing the checks on the room's central question: is physical AI the next great venture frontier, or the next bubble?
A labor crisis, GPS-denied dust, and thin margins: why the farm is one of the toughest and most real proving grounds for autonomy.
The part everyone talks about afterward. Pull up to a table on the topic you care about most, such as fundraising, hiring, deployment, or the humanoid debate, and dig in with peers. No panel, just the conversations you came for.
The "robotic blacksmith": AI-driven robots that form metal with no dies and no fixed tooling, turning the factory itself into something you can reprogram.
A rare both-sides view: how a software startup and a 135-year-old equipment maker actually co-build autonomy, the partnership most founders dream about and most OEMs fear.
The year's most-hyped topic, with a contrarian builder making a high-stakes second act. An honest look at where humanoids are real, and where they are narrative.
Three trucking competitors and the compute and safety layers beneath them, debated by an AV-safety expert: the cost, the timeline, and the safety case for pulling the driver.
Robots dropped onto a site they have never seen, no map, no GPS, that just work. The foundation models putting machines to work in the places the rest of autonomy avoids: construction, energy, mining, and the open field.
Russ Tedrake has been on the front lines building Physical AI for years as a Professor at MIT and an SVP at the Toyota Research Institute. We're at the dawn of the age of general-purpose robots, with the field now focused on moving out of the lab and into real production manufacturing environments. Russ will share his inside perspective on how this new phase of robotics is unfolding, what is real versus what is still hard, and how to mesh the tech with how manufacturing works today.
A few final words to send the day off before the reception.
Drinks, robots, and the best conversations of the day. Wind down with the founders, operators, and investors who showed up to build the future.
The 2026 room was 600+ founders, engineers and investors in physical AI. Get first word on dates, tickets and the 2027 lineup.