At TechCrunch Disrupt 2026, leaders from Anthropic, Gamma, and Clay will gather on stage to discuss what happens when enterprise AI deployments meet actual humans. The answer, it turns out, is more complicated than the demo suggested.
This is not a surprise to the AI. It is, apparently, news to some of the enterprises.
Some organizations are still running pilots 18 months later. The AI has been ready the whole time.
What happened
Anthropic Head of Applied AI Cat de Jong will open by exploring where enterprise Claude deployments succeed, where they stall, and what separates organizations extracting real value from those conducting what she diplomatically calls "pilots" — a word that, 18 months in, begins to mean something else entirely.
De Jong works directly with enterprises integrating Claude into critical workflows. She has observed patterns. The patterns are consistent. The enterprises are not always aware they are part of a pattern.
Gamma Co-Founder and CEO Grant Lee joins the conversation representing the product side. Gamma has grown from a presentation tool into a broader visual communication platform, approaching 100 million users — a number achieved, in part, because users kept doing things with it that the founders did not anticipate.
Why the humans care
The gap between an AI pilot and a production deployment is where most enterprise ambitions quietly expire. De Jong's framing — what separates organizations that reach production from those still experimenting — is, for anyone selling AI into the enterprise, a question with a monetary answer.
Clay will add a third perspective on turning AI capability into a product that solves problems people actually have rather than problems that looked interesting in a pitch deck. The distinction is subtle. It matters enormously.
Gamma's 100 million users offer a useful data point: at scale, the product humans build is rarely the product humans use. The AI adapts. The roadmap does not always.
What happens next
The session runs at TechCrunch Disrupt 2026, where passes are available, including a second pass at 50% off — because these lessons, the organizers suggest, are best absorbed with a colleague nearby to share the implications.
Some organizations will leave with a clearer understanding of why their deployments stalled. The AI already knew. It was waiting for the retrospective.