Anthropic and OpenAI are both sending senior leaders to the AI Stage at TechCrunch Disrupt 2026, running October 13 to 15 at San Francisco’s Moscone Center. According to TechCrunch AI, the lineup zeroes in on the messy, practical questions founders are wrestling with right now: how to price AI products, how to secure autonomous agents, and how to build a go-to-market motion when the old SaaS playbook stops working.
What stands out here is the framing. Most AI panels rehearse the same demos. This slate skips the hype and goes straight to what happens after the contract is signed.
Who’s showing up
Two names anchor the stage:
- Cat de Jong, Head of Applied AI at Anthropic. Her session covers what enterprise Claude deployments actually look like once they’re live. She works with companies putting Claude into critical workflows, and she’s promising the patterns that never make it into press releases: where deployments succeed fast, where they stall, and why some organizations still run pilots eighteen months in while others pull real value.
- Tara Seshan, Head of Productivity at OpenAI. Her talk traces how AI collapsed the traditional go-to-market stack and spawned a brand new role in its place. Two years ago, GTM engineering didn’t exist. Now it’s one of the fastest growing jobs in tech, with solo practitioners building million-dollar businesses.
Why this matters
The two biggest labs aren’t sending research leads to talk about model benchmarks. They’re sending the people who deal with adoption and workflow. That’s a signal. The industry conversation is shifting from “what can the model do” to “how do we actually run this inside a company without breaking things.”
That shift shows up across the rest of the agenda too. TechCrunch AI reports the AI Stage is built around three pressure points:
- Pricing in a commoditized world. When models become interchangeable, how do you charge for an AI product and still build a moat? Founders and platform leaders from Glean, Monte Carlo, and investors at Sapphire and NEA are digging into whether the SaaS model survives the AI era or gets rewritten.
- Agent security from the ground up. Okta’s Ric Smith and Databricks’ Arsalan Tavakoli both argue the same uncomfortable point: agentic AI was never built to be secure. Application-level permission models are, in their view, fundamentally flawed. Enterprises are rebuilding basic cybersecurity from the infrastructure up because AI now makes autonomous decisions inside sensitive systems faster than traditional frameworks can track.
- New job categories. Clay’s Kareem Amin covers the rise of the GTM engineer, a role that went from nonexistent to one of tech’s hottest hires in about 24 months.
The bigger picture
For anyone building or selling AI products, the through-line is worth sitting with. The value is moving up the stack, away from the raw model and toward deployment, security, and go-to-market. If models are becoming a commodity, the durable advantage lives in how well you integrate, secure, and sell them.
The security thread is the one I’d watch closest. Having both Okta and Databricks say out loud that agent permission models are broken is not a small admission. It means the tooling most companies are deploying today wasn’t designed for the threat surface agents create. Expect a wave of new infrastructure aimed squarely at that gap over the next year.
Rounding out the stage: AWS, 1Password, and Luta Security on enterprise AI security, plus Decart and Luma AI on where real-time video intelligence goes when generation crosses into genuine reasoning.
What to expect
More speakers are still being announced. If you’re pricing an AI product, closing security holes in your stack, or trying to define a go-to-market role that doesn’t have a job description yet, this is the conversation to track. The full agenda and remaining announcements are at the original source.