I keep ending up at the same whiteboard. A founder and I, two or three hours in, pulling their vision apart. Not because it's wrong — because it's too big to ship whole. The technology works. The demo is compelling. And the distance between what this thing could become and what the world will tolerate right now is making everyone in the room anxious.
You can feel it. The founder keeps gesturing at the future. Someone else keeps pulling back to the next eighteen months. Both are right. That's what makes it hard.
That anxiety is information. Two versions of the company pulling on each other — where this could go and what the world will underwrite right now — and the tension between them is trying to tell you something. The question isn't which one wins. It's whether the friction between them is producing anything — an opinion, a product, a direction neither side could have generated alone.
Most teams either collapse the tension — pick a shape and commit — or get stuck in it, the same argument circling week after week. There's a third option. Stay inside the friction long enough for it to produce something. Shape work starts when you stop choosing between them and start asking what each one is telling you.
I've watched teams argue about product direction for months. The same conversation circling, each person reacting to a different imagined version of what the product should become. Then someone puts a prototype in the room and the argument changes in minutes. Not because the prototype is right — usually it isn't. Because it creates a shared felt experience that the room can react to together.
This is why I build prototypes for portfolio companies — not to show them the answer, but to put something in the room that can push back. A rough, provocative thing that shifts the conversation from "what should we build?" to "what do we think about this?"
The prototype didn't resolve the argument. It turned abstract disagreement into shared material. Before: everyone reacting to their own imagined shape, their own private version of the future. After: a common surface to push against. That's what productive friction looks like. The output wasn't an answer. It was a shared object that made the next conversation possible.
I was in a room where the imagined shape and the measured shape had been arguing for weeks. The founder believed in the product. The dashboard said nobody else did. Same two shapes pressing against each other, wearing grooves but not moving.
I asked when the last time anyone had talked to the three users doing something unexpected with the product. Silence. The enacted shape — what was actually happening in the wild — hadn't been in the room at all.
A week later, the conversation was different. Not because the old tension resolved. Because a new tension replaced it — the enacted shape pressing against the imagined shape. The question shifted from "is the product working?" to "what are these users actually doing with it?"
You can feel the difference. When friction is doing its job, the room gets quieter. Someone says something they haven't said before. Positions move. When it's not, you hear the same arguments gaining volume. People get more certain, not less.
The move wasn't reducing friction — it was changing what the room pushed against. The old friction wore grooves. The new friction generated signal. A better question replaced a stuck one.
A portfolio company I've written about in this series — the lakehouse founders — had this exact gap. The imagined shape: every ai agent with its own isolated compute environment. Real, important, probably two years from the mainstream buyer. The ruled shape: capital that needed something underwritable in eighteen months.
It took weeks of midnight threads and wrong-but-useful prototypes. What emerged was a near-term product — isolation shipped to teams already wrestling with multi-tenant workloads. Not the old cost play. Not the full vision. A smaller, sharper shape the constraint forced into existence — a path through the present that kept the deeper architecture alive.
This isn't compromise. Compromise averages two shapes into something that satisfies neither. The slice lets the near-term shape ship, earn the right to exist, buy time — while the deeper shape stays protected.
The constraint didn't just produce a near-term product. It produced an opinion — a specific, hard-won position about what to ship now that the founder would never have developed without the pressure. That opinion is the kind of thing that can't be replicated by anyone who skipped the friction.
I've done this myself. Held the imagined shape as the lead in a board meeting where the enacted shape was saying something different — because the story I was telling investors depended on the old shape being right. I could feel the dissonance. I stayed with the story anyway. It cost us months.
The friction was there. I chose not to work it.
I've watched teams hand the weight to the measured shape too early, let the metrics decide before the product found its form. They optimized their way into a corner. These are the ways friction gets wasted.
Look at what each of those rooms actually produced. Shared material. A better question. An opinion. None of them are answers. But they're not nothing, either. And none of them are consensus.
I used to think opinionated products came from opinionated founders. Strong vision in, strong product out. But watching these rooms, I don't think that's right. The founder who shipped the near-term product didn't walk in with a position on what to build first. That position got produced — by weeks of constraint pressing against vision until something crystallized that neither side carried in alone. The opinions aren't the input. They're the output.
So the thing that makes a company hard to copy isn't what it knows. It's what it argued about. The specific frictions. The specific late-night threads where nobody could agree. That's the raw material. Arguments feel like waste — the thing you endure so you can get to the real work. But I think they are the real work. That's where the taste comes from.
In the first post in this series, I wrote about taste. Not an innate gift — something assembled, slowly, from watching what holds and what cracks. I think this is what assembles it: specific friction. Specific tensions between specific people. Those opinions accumulate. They compound. And at some point the company has a way of seeing that's genuinely its own. You can't copy that. You had to be there.
But here's what I keep sitting with. Taste isn't something you have. It's something that's happening. It's being produced, right now, in whatever room you're in — or it isn't.
This is why ai doesn't scare me the way it scares some people. When code and research and market maps all compress toward zero cost, the question isn't whether your company has taste. It's whether your company is still producing it. A team that resolved all its tensions, that found its shape and settled in — that team's taste is already dying. Not because they forgot anything. Because the friction that was making it went quiet.
I keep returning to the same image. The whiteboard. The anxiety. The founder gesturing at the future while someone pulls back to the present.
The room isn't a problem to escape. It's where taste gets made.
The moment it feels settled, something shifts. A new customer. A market turn. A founder who finally sees what they've been building.