Founders wiring agents into their stacks need to know whether the product is actually working before they scale the automation — Rahul Vohra's PMF method gives you the measurement apparatus to find out. The hard part is still yours: deciding what the number means and who to listen to.
Vohra's framework, published in First Round Review, starts with a deceptively simple survey question: how disappointed would you be if this product disappeared? Once you hit forty-plus responses, you segment ruthlessly — isolate the people who answer "very disappointed," understand who they are, and let that cohort define your real user. Then you split the roadmap: one lane doubles down on whatever those lovers already value, the other lane addresses the specific objections raised by people who said "somewhat disappointed." The target is forty percent very disappointed. Below that, you iterate. Above it, you push.
When agents handle onboarding, support, or core workflows, the PMF signal gets noisier fast. Users form impressions of your product without ever fully understanding what fired under the hood, which means the "very disappointed" cohort might be loving a behavior that was half-accidental and hard to reproduce reliably. The Vohra method holds — run the survey, find the forty percent, segment hard — but the founder's job is to trace backward from what those users love to the actual agent behavior producing it, then make that behavior intentional and durable before scaling it.
- Run the survey before you automate further, not after
- segment your "very disappointed" users to find the agent behavior worth locking in
- treat "somewhat disappointed" feedback as a design brief for where human judgment in the loop is still missing.
