PMF Is Gonna Rock Ya!

This post is the written version of my breakout session, PMF Is Gonna Rock Ya!, at the Northeast Entrepreneurship Summit, hosted by The Eleven Society, on Sunday, November 15.

I’m also judging the startup competition when eleven schools pitch at the summit. The session is intentionally small and informal: a conversation for liberal arts undergrads from NESCAC schools, not a lecture. More on The Eleven Society.

“If you want to do something, go and do it! Don’t wait to be asked…” — Bill Drummond, The KLF, The Manual (1988)

That’s how two men following their own recipe of “little money or talent” — an ex-label A&R man and a squat-dwelling guitarist — put five singles in the UK top ten. Bill Drummond and Jimmy Cauty, better known as the KLF (also the Justified Ancients of Mu Mu and the Timelords), were not virtuosos — conceptualists who sampled and collided other people’s records. In 1991 they were the biggest-selling singles act in the world.

Silicon Valley has been trying to write the same book for decades. We call it product-market fit. PMF. Does the product work, does anyone want it, and do they want it enough to use it, pay for it, and keep coming back?

You don’t need to invent a new technology. Sometimes the opportunity is recognizing what existing technology makes possible.

Here are five examples from companies we’ve invested in at PWV (Preston-Werner Ventures).

Doctorin’ the Product / Last Train to Model Central

Cursor: change the product

We first looked at Cursor at its seed round. Frankly, the product and developer experience — the DX — still needed work. We liked the founders, and we believed AI would fundamentally change how software gets written, but the product wasn’t there yet, and large incumbents like GitHub were already moving on the same opportunity.

This was just as GitHub Copilot was arriving. GitHub had access to more of the world’s code than anyone, and we speculated they would use that advantage to own the coming explosion in what we now call vibe coding. They didn’t. The Cursor team grabbed the moment instead.

Only a few months later, at the Series A, we saw it again, and it was dramatically better. Same founders. Same thesis. A very improved product.

The four founders had spent about six months before founding Anysphere building an AI tool for mechanical engineers working in CAD. None of them knew the domain, and it went nowhere. They pivoted to code — Cursor shipped in 2023 — but they’d already spent considerable time figuring out what to build.

And even after finding the right direction, they kept improving it.

The lesson: your first idea is a hypothesis, not your destiny.

Question: what’s the hypothesis you’re currently treating as destiny — and what would it take to test it this month?

fal: change the market

Where Cursor changed the product, fal changed the market.

Burkay and Gorkem started by building Python data infrastructure for data pipelines, analytics, and ML training. It was highly optimized, but it was not the business that made them successful.

We had been looking for a platform to generate images, video, and music. Most demos were slow or clumsy. Then we saw fal’s “Make it Real” collaboration with our portfolio company tldraw: you sketched on screen, and an image appeared milliseconds later.

That was the tell. Their infrastructure was exceptionally good at running generative models at scale. When Stable Diffusion took off, they pointed it at the new market.

When we first met the fal founders, they told us the future is one where every pixel will be “generated, not rendered.”

fal has made that happen. They saw the market before it was obvious — every product would eventually need fast, cheap inference for generated media — and they built for it. That is product-market fit: the market showed up, and they were already standing in it.

Neither started with exactly the business that made it successful. Neither had to invent all the underlying technology. They recognized an opportunity, made something useful out of it, and followed the evidence.

Sometimes the best opportunity isn’t the one you set out to find.

Question: what infrastructure or skill you’ve already built could you point at a completely different market?

What Time Is Fit?

Turso: the market catches up

I consulted with Supabase back in 2021 and saw first hand how Supabase won the previous era by being an all-in-one platform: Postgres, auth, storage, edge functions, and a heavy Docker setup. Training data and vibe-coding tools like Lovable defaulted to it.

Turso built for unlimited SQLite databases: one per user, one per agent, one per task. Its multi-tenant isolation and local-first experience were already suited for the agentic era.

PWV had already invested in Turso’s angel and seed rounds. Glauber believed the future was millions of small databases. At the time, that was an emerging idea, not yet a market.

Then AI agents arrived. They need to spin up a database as easily as creating a file, with just as little concern about cost. No provisioning a dedicated machine for every small workload. Databases cheap to create, available on demand. Turso had the right shape for that.

That’s why, in October 2026, the company that won the last era came to acquire the company built for the next one.

Sometimes you don’t find the market. The market finds you.

Question: what’s a bet you’re making on where the market will be in two years — and how will you know if you’re early or just wrong?

Workbrew: find the buyer

Developers love Homebrew, the free tool for installing software on macOS and Linux. Corporate IT departments do not love employees installing whatever they want. They need visibility, security, and control.

Workbrew built a business in that gap. Developers use the software. Companies pay to manage it.

The person who uses your product isn’t always the person who pays for it. This is really about understanding people.

Two groups can look at exactly the same product and see completely different problems. Recognizing those differences is often where the business opportunity lies.

Question: what’s a free tool developers use that companies would pay to bring under control — and what would each side say they’re buying?

Justified & Validated?

Zoo: become the layer

Zoo is another PWV investment, and it shows a different way to take on an entrenched market.

CAD has been one of the most entrenched software markets in the world for over 50 years: proprietary file formats, workstation-bound execution, and workflows never designed for cloud compute or machine learning. Hardware design data mostly sits in static files that are hard to use as training data.

Zoo rebuilt the CAD stack from first principles: KCL turns design intent into machine-readable code, and a GPU-backed geometry engine runs through an API. It doesn’t have to replace the incumbents — it plugs into existing stacks, connecting design, analysis, and manufacturing. Blue Origin is already feeding millions of CAD files into a fine-tuned Text-to-CAD model — early proof Zoo can expand the market, not just compete inside it.

The analogy to Cursor is direct: rebuild the workflow around the model rather than bolt AI onto the old stack.

The lesson: you do not have to replace the incumbent to win the market. Become the layer it cannot build itself.

Question: what’s one deeply entrenched market you would never try to replace — and what layer could you build underneath it?

It’s Grim Pre-Fit / Burn the Runway

Five companies. Five different lessons.

Company	PMF lesson	What it requires
Cursor	Change the product	Question assumptions
fal	Change the market	Recognize opportunities
Turso	The market catches up	Have a point of view
Workbrew	Find the buyer	Understand people
Zoo	Become the layer	Connect the lifecycle

There’s no single formula for product-market fit. Much of what makes great founders successful has little to do with being the best programmer in the room: curiosity, creativity, taste, critical thinking, understanding people, and the willingness to change your mind. Those aren’t exclusive to liberal arts graduates, but they’re exactly what a liberal arts education develops.

You don’t need anyone’s permission to recognize a problem worth solving, and you don’t have to invent the underlying technology. You can learn the tools, find collaborators, or bring together complementary skills. What matters is having a point of view about how the world could be different, and being willing to find out whether you’re right.

Which brings me back to The KLF. They didn’t invent the instruments. They took what already existed, combined it in original ways, and made a mark on popular culture. They found an audience, became enormously successful, and even wrote the manual.

Then, in 1992, they quit the music business. Two years later, as the K Foundation, they took a million pounds in cash to a Scottish island called Jura and burned it. Literally.

As a venture capitalist, I’ve seen plenty of startups burn through a million pounds, but rarely quite so efficiently. They framed it as a deliberate artistic act, not a business failure. And it makes one last point:

Fit isn’t a finish line. You find it, lose it, re-earn it — and one day, if you’re lucky, you build something that outlasts you.

The KLF had the creativity, taste, and point of view to make something people wanted. That’s the connection I want to leave you with: you don’t have to invent the instruments to change the music. You need an idea, the imagination to do something with it, and enough curiosity to discover whether anyone else cares.

---

Tools: fal for image generation (model google/nano-banana-2.1). Command Code v1.88.5 for writing and editing (model deepseek-v4.1-flash-fast-(latest)).