Product market fit is a verb.

You don't arrive at product-market fit like a destination on a map. PMF is rented and you are always just on the edge of eviction.

Startups exist along a spectrum between two opposing vectors. One is a decay function. Every tick of the clock, along one dimension or another, your business is decaying. The other is rate of growth. The degree the latter compounds and the former is minimized is the fulcrum upon which success hinges.

Everyone in startups knows this, but it is so much more true in the age of AI that if you really want to succeed, you should probably tattoo it on your chest. Only the paranoid survive.

Product velocity is the single most important metric for engineering and product orgs. Roughly speaking, product velocity is the time it takes to deliver meaningful experiences to your customers.

Here is why it is so much more important now more than ever.

Proposition Number 1:

AI is commoditizing programming. As the barrier to entry lowers, the result will be a Cambrian explosion of vendors and tooling.

Proposition Number 2:

AI is changing what software is. It is changing both what is possible, and redefining legacy consumer and computing interfaces. The rate of AI progression itself is so dramatic that nearly every week the range of what is possible, and even optimal, expands. When OpenAI released its structured output endpoint, an entirely new class of application could be built. Similarly with its audio, image and video models.

The conclusions that follow are:

(1) Means a dramatically increased competitive landscape, not only from current and future vendors but also from in-house software teams. Every vendor that sells horizontally across markets will face a thousand competitors with bespoke vertical applications. The increased competition will lead to pricing wars.

(2) Means the ground itself is shifting so rapidly that you might be obsolete even before the ink is dry on your first customer contract. Consumer preferences will trend toward what is optimal and if what is optimal is itself changing as AI expands what is possible, last month's innovation may be this month's dead weight. The rate of decay has increased by an order of magnitude.

Your product velocity has to be (1) faster than the current competition (which itself is more varied, less encumbered, and stronger than it was before) but also (2) keep pace with the rate of AI innovation and consumer preference in general (which is faster than most fast teams ship). If you do (1) and not (2), you will be a victim of the Innovator's Dilemma, though on a much shorter timeline than we've historically been used to.

How to radically increase your product velocity?

Be wary of sunk cost fallacy and any old world PDLC process or bottleneck. Delete everything. If you don't have to add 20% back in, you did not delete enough. ‘

You should be using tools like Cursor and Windsurf to accelerate engineering. Your design system is likely a sunk cost, and if AI can't write it or maintain it, delete it. Every PM should be building prototypes hand-in-hand with customers and stakeholders to accelerate what to build and why questions using tools like v0, Lovable, or Bolt. You should be releasing multiple times a day using a Trunk Based CI/CD, rolling releases, and feature flags. An AI should review every PR and you should have an automated AI-first way to capture customer feedback that inks the next set of tickets and drives the next set of meaningful product improvements.

The result is a PDLC that will feel much more like a bottom-up emergent phenomenon than something gate-kept and top down. It will be much more like a tango than a waterfall. If you do this correctly and you're serious about, you might just avoid the coming extinction.