Jason Lemkin: AI Products With No Moat Today Can Build a Massive One in a Year
Watch on TikTok
Lemkin's argument is that moats in AI are earned through compounding product depth, so a product that is trivially clonable at launch can become nearly impossible to copy within a year. In this 28-second clip from the 20VC podcast, Jason Lemkin of SaaStr uses the vibe-coding wave of Bolt, Replit, and Lovable as evidence, then applies the same logic to Instinct, the AI personal assistant that jumped to a $2.5 billion valuation in weeks. The on-screen caption frames the question directly: could Instinct build the next big moat in AI?
Everything Was Built in a Month
Lemkin opens by recalling how the first wave of AI app builders launched. "When these products came out, they were all built in a month. Bolt, Replit, Lovable, and 22 others, Base 44, Base 56." The "Base 56" line is a joke about how interchangeable the field looked. His point is that cloning any of these products in the early days took little effort, and dozens of teams did exactly that. Speed to launch conferred no defensibility because everyone had the same speed.
Depth Is the Moat
The clip's core claim is that these products stopped being clonable once they accumulated serious capabilities. "Now they're so complicated. They're doing pen testing, security automation. These are such rich products." A competitor copying the surface of Replit or Lovable today would still be missing the security tooling, workflow depth, and operational maturity built over the past year. Lemkin states it plainly: "Replit and Lovable of a year ago and Bolt were not a moat. Today, they have massive moat." The moat came from shipping faster than anyone could copy, not from any structural advantage at launch.
The Bar for Instinct: 100x in a Year
Lemkin then sets the test for Instinct. "If Instinct's gonna do what we claim it does, in a year, it's gotta do a hundred times more than it does today. And that is still a moat." The demo footage in the lower panel of the video shows why the claim is plausible. One frame shows a text thread where Instinct messages a user: "Heads up: you're on AA148 today, SFO 3:20pm to JFK 11:59pm. I'm checking you in now and will send your boarding pass." The assistant handles the task end to end over SMS, with no app involved. Executing real-world tasks like this across travel, email, and scheduling is the kind of accumulating workflow depth Lemkin is describing.
What the Frames Show
The video is a split-screen podcast clip. Lemkin appears in the top panel wearing headphones, with host Harry Stebbings visible in the bottom panel in some frames. Interspersed demo footage includes an AI builder prompt box with the request "Add a sidebar with user profile stats summary and account," the Instinct flight check-in text exchange, and a growth counter reading 52,977 over an animated curve. The editing pairs each stage of Lemkin's argument with product evidence: easy-to-clone builder UIs early, then an agent doing real work, then traction numbers.
Why This Matters for AI Founders
The practical takeaway is that launch-day defensibility is the wrong question for AI products. The right question is whether the team can out-ship every cloner for a sustained period. Instinct fits the pattern: it launched invite-only, works through SMS and WhatsApp, and raised a $250 million Series B at a $2.5 billion valuation in August 2026. Coverage from TechCrunch and Forbes notes both the fundraising frenzy and the privacy questions that come with an agent holding that much access. Lemkin's bet is that if the product keeps compounding capability at that pace, the moat follows. "I think it could easily happen with an Instinct."
Key Takeaways
- Early AI app builders (Bolt, Replit, Lovable) were built in about a month and were easy to clone at launch.
- Those same products now have deep capabilities like pen testing and security automation, which makes them hard to copy.
- Lemkin's rule: a product with no moat today can build a massive one by shipping faster than competitors for a year.
- His bar for Instinct is doing one hundred times more in a year than it does today.
- The moat in AI is compounding product depth, not launch-day differentiation.
Resources
Published September 4, 2026. Writeup generated from a favorited TikTok.