TechCrunch Disrupt 2026 is loading its AI Stage with a lineup that skips the hype and goes straight for the operational realities founders are wrestling with. The conference runs October 13-15 at Moscone Center in San Francisco, with Google for Startups presenting, and the early pricing window that saves attendees up to $200 is about to close.
Anthropic’s Head of Applied AI, Cat de Jong, is doing something rare: talking about what happens after enterprises deploy Claude. Not the pilot phase. The actual deployment. She works directly with companies putting Claude into critical workflows and sees patterns that never make it into press releases — where things succeed immediately, where they stall, and what separates organizations extracting real value from those still running pilots eighteen months in. That’s a brutally specific timeframe, and it tells you how much patience the enterprise market has burned through.
OpenAI’s Tara Seshan will trace how AI collapsed the traditional go-to-market stack. Her framing is blunt: two years ago, GTM engineering didn’t exist. Today it’s one of the fastest-growing roles in the industry, with independent practitioners building million-dollar businesses. Clay co-founder Kareem Amin is covering the same ground from a founder’s perspective. The throughline is clear — AI didn’t just add a tool to the sales stack, it dissolved the stack entirely and reassembled it around something new.
Security gets two dedicated sessions, which says a lot about where enterprise anxiety sits. Okta President Ric Smith is arguing that application-level permission models are fundamentally flawed for agentic AI, and that agent security requires rebuilding cybersecurity infrastructure from the ground up. Databricks co-founder Arsalan Tavakoli is tackling observability and governance for AI making autonomous decisions inside sensitive systems. Notably, Tavakoli’s session title frames the problem as one of assumptions, not broken enterprises — a signal that the industry’s mental model, not just the technology, needs an overhaul. Across the stage, the honest thread is that AI has outpaced every framework built to govern it, and the people doing the rebuilding are the ones on these panels.
