Product-Market Fit on Fast-Forward: How AI Compresses the Incumbent Advantage

I’m working my way through Reforge’s AI Strategy course and so far it‘s sparked a lot of great thoughts! In particular, it’s been a forcing function to pressure-test my own instincts about how to build AI products and modify my approach to product strategy for the AI era. My next three posts are going to be an encapsulation of the major concepts that have stood out to me and how I’m going to be applying them in my day-to-day.

Two of the first ideas from the course that are shaping my mental model are: (1) the Product-Market Fit (PMF) treadmill, and (2) how AI’s rapid proliferation can lead to PMF collapse.

TLDR: Incumbents that once had years to adapt to technological change now get months (or weeks) before upstarts leveraging new, AI-first ways of doing things can supplant their market position. This makes it more important than ever to continually search for strong PMF, reassess PMF regularly, and not become apathetic about innovation.

What Is Product-Market Fit?

Product–market fit (PMF) is the point where a specific product consistently solves a valuable problem for a clearly defined audience so well that demand for the product becomes sustainable and efficient—customers adopt it, keep using it, and recommend it, with growth increasingly driven by pull rather than push.

Here are a few quick sanity checks to judge your product's PMF:

• Strong retention/engagement
• Efficient growth (organic/word-of-mouth share rises; sales cycles shorten; healthy unit economics)
• Clear segment–problem–solution alignment (you know exactly who it’s for and why they prefer it)
• Qualitative love (users say it’s a “must-have”; e.g., ≥40% would be “very disappointed” if it went away)

The PMF Treadmill Just Hit Fast-Forward

PMF is a critical aspect of any successful product. There’s an inclination to thinking that finding PMF is an exercise for early-stage products that, once attained, can’t be lost. Even prior to the chatGPT revolution, that was a fallacy.

In reality, finding PMF is more like a treadmill where you’re always striving for what’s next, rather than a milestone. The treadmill doesn’t operate at a steady pace—it’s continually speeding up in response to changing factors: customer expectations rise, competitors copy or leapfrog, and technological capabilities shift. As these speed changes happen, products need to adapt in order to retain product-market fit. Historically, major tech transitions (e.g., mobile) increased the treadmill’s speed at a manageable pace with capabilities that matured on slower timelines, ecosystems that formed over years, and user adoption that followed steadily after capabilities increased.

The current pace of AI innovation breaks that cadence:

• Capabilities update weekly
• Building skills are becoming democratized
• Distribution is instant and cheap
• User expectations spike in step changes, not slopes

The result of this changing dynamic is PMF collapse, where core product growth models that felt durable suddenly unravel as users discover faster, cheaper, more personalized alternatives. This can lead to overnight loss of PMF as users’ needs are better met in new, AI-first products.

What Collapse Looks Like

AI-induced PMF collapse has happened to many companies since the release of chatGPT. One poignant example of PMF collapse is Stack Overflow. Stack Overflow is an online community where engineers and other technical folks answer questions about code, technical design, and more. When I was a software engineer, consulting Stack Overflow was a very common feature of the job to answer questions about technologies you were considering or trying to debug issues.

Stack Overflow suffered a near-immediate, intense drop off in usage with the launch of Copilot, an AI tool that allowed developers to ask questions about code and was contextual to their code base. These tools have become even more popular since the launch of Copilot, with tools like Cursor, Claude Code, and even basic chat assistants offering personalization and zero wait for community members to respond.

In response to these new tools and the ability to get personalized responses, Stack Overflow lost PMF very quickly and is still trying to recover from that loss.

Why AI Is Different (and Less Forgiving)

Most technology shifts follow four steps: (1) capability emerges → (2) products ship → (3) distribution spreads → (4) expectations rise. This cadence takes place over a longer time horizon. With AI, those steps compress:

• Emergence → shipping: weeks.
• Shipping → distribution: days.
• Distribution → rising expectations: near immediate.

That compression eliminates the “breathing room” incumbents relied on to gather data, socialize a response, and execute a multi-quarter pivot. You either predict the hit and move early or get trapped by inaction.

Assessing Your Risk

Reforge proposes an 18-factor framework to assess a company’s risk of PMF collapse. The framework assesses PMF risk of collapse using a score for each factor 1 (low risk) to 7 (high risk) and encourages PMs and their teams to go through the framework and rank their product.

The first step to knowing where to go is knowing where you are. Having an honest conversation about the state of your products–the good, the bad, and everything in between–will put you on strong footing to begin doing something to shore up your strengths and neutralize your weaknesses.

From Diagnosis to Direction: What to Do Next

Once you have a good grip of your current state, the below items present some actions that may make sense to take.

1) Move to the Primary Workspace

If you’re an adjacent tool, embed where work happens: IDEs, docs, canvases, CRM, help desk. Ship native plugins, APIs, or a thin client that feels first-party to your users.

2) Bind to Proprietary Data

Make your AI uniquely good because of data only you can see: customer telemetry, domain-specific knowledge bases, private data. Your proprietary data (intelligently utilized) gives you a unique edge that no competitor could match.

3) Rewire Your Growth Loop

If your loop depended on human contribution, shift to AI-assisted creation models with human curation. Double down on the importance of creating strong communities. Build shareable artifacts and collaboration surfaces where AI amplifies reasons to contribute.

4) Align Pricing with Outcomes

Phase in value metrics (resolutions, transactions, usage tiers) and safety rails (caps, credits). AI costs per transaction and, as such, it’s important to factor in the cost of AI into your pricing models. Keep one simple plan for procurement and one outcome-aligned plan for AI-heavy teams.

5) Fortify Switching Costs (Ethically)

Deepen workflow integration and drive customer benefits from increased usage of your tools (context improves over time, model tunes to your org). Offer migration accelerators so you are the easy switch from competitors.

6) Decide Your Speed: Incubate vs. Cannibalize

Stand up an AI “speed lane” team that can ship in weeks, not quarters. Give the team scope to cannabalize your legacy business where needed. In AI shifts, the product that eats you should be yours!

7) Curate Evals

As you build new AI features, focus on curating strong evals to keep a pulse on the performance of your AI tools and help you tune outputs. I’ll do another post about making great eval frameworks, but suffice it to say they are extremely important to the success of AI products.

The Hard Truth (and the Opportunity)

High vulnerability to PMF collapse isn’t the worst place to be—inaction is. The companies that navigate tech shifts predict the hit early and bet the company on the right transformation. Use the above scorecard to pick your leverage points, move work to where users live, fuse AI with proprietary data, and realign your growth loop and pricing to the new value you’re driving.

The treadmill isn’t stopping. But with the right map and cadence, you can keep pace!

To read more about these ideas, check out Reforge’s post on PMF Collapse and the PMF Treadmill int eh age of AI!