PMF can be lost suddenlyWhen Product-Market Fit Collapses

PMF is a state, not a destination

The conventional narrative around product-market fit treats it as something you achieve and then keep. Find PMF, scale on top of it, defend the moat.

Brian Balfour's argument in Product-Market Fit Collapse^[1] is more uncomfortable: PMF is a state, not a destination. It can be lost. It can be lost suddenly. And AI is causing it to be lost in categories that previously felt stable.

The collapse pattern is specific. It's not slow decay — that's the normal lifecycle of any product. It's sudden, large-magnitude loss of fit because the market itself has moved in a way that makes the product's value proposition obsolete.

Homework-tutoring services lost most of their PMF in months when ChatGPT made the core use case zero-cost. Copywriting agencies built on $0.10/word freelancers face the same dynamic. Sub-$10/month productivity SaaS products face the question of whether AI-native alternatives have made their core feature set table-stakes.

Why AI accelerates collapse

Previous technology shifts caused gradual displacement. Incumbents had time to adapt because adoption was slow, expensive, or required infrastructure. AI is different in three ways that accelerate collapse:^[1]

1. Adoption is fast. No installation, no procurement cycle. A user can switch from a paid SaaS tool to a ChatGPT prompt in seconds.
2. Cost is near-zero at consumer scale. Free tiers across major models. Most consumers and many small businesses can replace paid tools with no marginal cost.
3. Capability ramps fast. Models improved roughly 10x in capability between 2022 and 2025. Use cases nonviable in 2022 became dominant in 2024.

The combination produces collapse curves steeper than anything previously seen in software.

The four-factor risk diagnostic

Reforge's framework^[2] offers a four-factor diagnostic:

• Use Case risk. Can the core user task be done with a general-purpose AI tool? Homework: yes. Brain surgery: no.
• Growth Model risk. Does your acquisition depend on channels that AI is reshaping? SEO content sites face AI-Overview risk. Paid social faces creative-cost compression.
• Defensibility risk. What makes you hard to replace? Brand, integrations, data moats, network effects — or just a UI on top of capability that's now commoditized?
• Business Model risk. Does your monetization assume features that AI is making free? Per-seat subscription on a capability that's now consumer-grade?

High risk on any one factor is manageable. High risk on three or four factors is collapse exposure.

What to do about it

The Reforge prescription^[1] is uncomfortable: lean in. Don't defend. Build the AI-native version. Don't wait for collapse — accept that current PMF is decaying and build the next state actively.

Reforge itself did this. They shipped five AI-native products with a team of 20, including their AI prototyping tool. The bet wasn't "AI will displace some of our content business." It was "AI will displace all of it unless we build the AI-native education company first."