Shield AI, Waabi and GM to discuss building AI when failure isn't an option

A panel at TechCrunch Disrupt 2026 will tackle the gap between laboratory models and systems that must survive in the physical world. The session, "Building AI Systems When Failure Is Not an Option," runs on the Real World AI stage. Nathan Michael of Shield AI, Raquel Urtasun of Waabi, and Mikell Taylor of General Motors will focus on safety culture, validation, regulatory hurdles and trust — not accuracy benchmarks.

Michael, Shield AI's chief technology officer, leads development of Hivemind, platform-agnostic autonomy software built for environments where a mistake brings down an aircraft. His background includes years at Carnegie Mellon's Robotics Institute, where he directed the Resilient Intelligent Systems Lab. In February, Hivemind was selected as the autonomy provider for the U.S. Air Force's Collaborative Combat Aircraft prototype program. A month later, the company raised $1.5 billion in a Series G round at a $12.7 billion post-money valuation. That experience positions him to address how to verify a system is ready for operational missions.

Urtasun brings 25 years in AI and autonomous vehicles. Before founding Waabi, she served as chief scientist and head of R&D at Uber ATG. She is a computer science professor at the University of Toronto and a co-founder of the Vector Institute for Artificial Intelligence, with more than 200 publications. In January, Waabi raised $1 billion and announced a partnership with Uber to deploy 25,000 robotaxis powered by the Waabi Driver. The core of the approach is Waabi World, a simulator that trains, tests and stresses the virtual driver in a synthetic environment. Urtasun herself has noted that the company's autonomous trucks still require full validation before driverless deployment, making the readiness question especially central for her.

Taylor now leads robotics strategy at GM's Autonomous Robotics Center after two decades building robots for practical tasks. At Amazon Robotics, she led the team that developed Proteus, the company's first fully autonomous mobile robot. Her career began with a less industrial project — a prom-date robot — but has since focused on systems meant to function in human environments. Her perspective completes the picture: not just aircraft or trucks, but robots that share workspace with people and demand an entirely different safety standard.

Organizers note that the discount window for Disrupt passes — up to $200 off — closes tomorrow, September 25, at 23:59 Pacific time. Prices rise after that. For anyone building or researching AI systems that touch the physical world, the session brings together three viewpoints that rarely share a stage — military, transportation and industrial — around the same fundamental problem: how to know a system can be trusted once it leaves the lab.