Only at TechCrunch Disrupt 2026: What happens when OpenAI ships your roadmap?
Sep 14, 2026, 8:00 AM · TechCrunch

TechCrunch’s Disrupt session frames the real AI-startup threat: not another founder, but OpenAI, Anthropic, or Google shipping your feature as a platform update.
Why it matters
For years the competitive map for AI startups was peer-to-peer. That map is broken. Every major foundation-model release now forces the same board question: what happens if the thing you spent a year building becomes a checkbox in the platform?
TechCrunch is putting that question on the Builders Stage at Disrupt 2026 (October 13–15, Moscone West), with Michel Tricot of Airbyte, Linda Tong of Webflow, and Rob Toews of Radical Ventures. It’s a promotional write-up for a conference session—but the strategic anxiety it names is already reshaping product plans, fundraising decks, and valuations.
From the desk
We’re not covering this as event marketing. We’re covering it because the question is the story.
When the platform can absorb your feature, “Can we build it?” stops being the hard part. “Can we still own it?” takes over. The session’s answer—and ours—is that lasting companies will sit where models are weak replacements: proprietary data, embedded workflows, customer relationships, domain expertise, and trust. That’s not romantic. It’s the only durable inventory left when capability keeps getting commoditized.
We’re for useful AI products that survive the next model drop. A thin wrapper that dies when GPT ships the same button was never a business; it was a temporary arbitrage. The harm if founders ignore this is capital burned on features that become free, employees hired into products that evaporate, and customers stranded when the startup becomes a demo.
Tricot’s Airbyte angle matters: infrastructure underneath analytics and AI, with thousands of customers including a meaningful Fortune 500 slice, is the kind of glue platforms don’t casually rewrite. Tong’s Webflow view is the SaaS survival problem—evolve the product without surrendering the moat. Toews is the investor filter: what still looks like a company in three years?
I’m watching whether founders leave Disrupt talking about workflows and data ownership—or still pitching “ChatGPT but for X” as if the platform won’t notice.
Context
Disrupt is pitched as TechCrunch’s flagship gathering of founders, investors, and operators. Early-bird framing in the piece notes pass savings before a September 25 rate increase. The underlying competitive shift does not depend on attending the session.
Who feels it
- AI startup founders
- Roadmaps need a platform-absorption test: what still matters after the next OpenAI or Anthropic release.
- Investors
- Defensibility diligence shifts toward data, workflow lock-in, and switching costs—not model novelty alone.
- Enterprise buyers
- They should prefer vendors whose value survives vendor model upgrades, not ones that vanish when the platform copies the feature.
What to watch
- Concrete examples from the session of startups that survived platform encroachment—and ones that didn’t.
- Whether fundraising language moves from model quality to proprietary workflow and data claims.
- Foundation-model product surfaces that quietly absorb vertical features founders still treat as differentiation.
Companies: OpenAI