It started on a Saturday. Claude Code open, no plan, just a question I couldn't stop turning over from a call earlier that week — about the shape a technology should take when it meets the world. Not the code. The container. Who encounters it first, what it looks like when they do. Four hours later I had a mock so rough it wouldn't have survived a code review. But it was a thing you could stand inside — a sketch of a possible future, just tangible enough to react to.
I keep reaching for this move. When a conversation has been circling — when everyone in the room is smart and engaged and still somehow stuck — sometimes the only thing that shifts it is an object. Something the room can gather around and push against.
I've started calling it shape injection. And it keeps surprising me — not just with what it does in the room, but with what it's taught me about shape. Shapes move. Faster than I expected.
A portfolio company. Data infrastructure — strong engineering, real technology, developers who loved the product. The kind of company where the technical foundation is so solid it becomes its own gravity. The team thought in schemas and query performance. The made shape — what they'd actually built — was elegant, powerful, and entirely oriented toward people who think in code.
But there were users at the edges — governance teams, compliance officers, business analysts — who needed to ask questions of their data that had nothing to do with implementation. Does this include pii? How was this credit score calculated at this moment in time? What changed since last quarter? These weren't developer questions. They were the questions that keep someone up at night who never opens a terminal.
The imagined shape — the team's picture of what the product was and who it was for — hadn't caught up. The founder's years of building for developers kept pulling every conversation back to the architecture, the performance benchmarks, the technical elegance. And the enacted shape — what users were actually doing — wasn't loud enough yet to make anyone look up.
I tried words. They bounced. Not because anyone disagreed — because the thing I was pointing at lived in felt shape — in bodily experience, the gut sense of using something — and that doesn't travel through conversation.
So one weekend I built two things. Both rough, both meant to be argued with.
The first was a mock of a control plane oriented toward governance. Compliance workflows. Data lineage visualized for someone who doesn't read sql. Barely functional, but it tried to answer the question: what would it feel like to be a compliance officer opening this product on a Monday morning?
The second was an agent prototype, built on top of their own framework, that answered business questions in plain language. The kind of question a cfo asks at four o'clock on a Friday when something doesn't add up. The agent used their technology. It just spoke a different language.
When the team saw the mock, something unexpected happened. Nobody talked about the design. They talked about what it implied — including things I hadn't built it to say. The mock had become its own object in the room, pulling the conversation somewhere none of us had pointed it.
The governance framing shifted the competitive landscape. The product wasn't measured against tools with a decade of head start on query performance anymore — it was doing something those tools had never been built to do. And the developer-first story the team had been telling started to share space with a different story, one that included the people at the edges. That's the part nobody mentions about reframing — every new shape also draws a quiet line around who the product isn't for. At least not yet.
The agent prototype did something similar. Watching an ai answer a business question using their own infrastructure — a business user asking, not a developer — made something possible that no pitch could have made real.
But the shift didn't happen in that room. It seeded. Slowly, over weeks, sometimes surfacing in offhand references — a faint echo in how the founder described the product to someone new. They didn't adopt the governance framing. They metabolized it. What they eventually built was different from what I'd mocked, and it should have been.
Shape injection doesn't produce a decision. It produces a drift — a new attractor the team didn't have before. You can't control where it goes.
I've been trying this with multiple artifacts — three or four rough prototypes, each one a different future the company could inhabit. It started as an experiment and it keeps working in ways I don't fully understand. When there are several futures to react to, people stop debating abstractions and start comparing felt experiences. The stuck loop breaks because the object in the room keeps changing.
Another portfolio company — one I've written about before in this series. We'd spent months working a shape transition together. The team had converged on a direction that dissolved old dependencies and opened new territory. It felt like the room had found something. A real shape.
Then the founders saw something else. Not from the direction we'd agreed on, but from a broader pattern they could read because they were closer to the market than anyone. A different future. Bigger. More aligned with what they'd built before but touching new territory.
In an earlier version of this company — before anyone had vocabulary for what was happening — this might have played out as confusion. A pivot nobody agreed on. A board conversation where someone asks "are we still doing X?" and the room goes quiet.
Instead, the team could name what was happening. The new imagined shape was compelling, but the measured shape — traction, data, evidence — was absent. Nothing to anchor the excitement to. We have an imagined shape that's pulling us, but the measured and enacted shapes haven't caught up. The founder knew what the gap was. The work was clear: go build evidence. Go make this real enough to evaluate.
And the old shape — the one the room had converged on months earlier? I found myself arguing to keep it alive. Not as the primary direction, but as a coexisting tension. Because collapsing it would lose information the team might need when the world shifts again.
This is what shapes do. They move.
I keep coming back to this. The artifact I built on a Saturday didn't produce a decision — it seeded a drift. Something shifted in how the founder talked about the product, and it kept shifting over weeks. The mock was static. What it provoked wasn't. And that gap — between the artifact's fixity and the response's movement — pointed at something I'd been slow to see.
Most product conversations treat shape as a decision. We're a developer tool. We're an enterprise platform. We're an ai-native database. As if you pick a shape, execute against it, and the work is to hold course. But I don't think shape is a decision. I think shape is a tempo. It has speed, direction, and drift. And the shapes I named earlier in this series — imagined, founder, made, measured, felt, ruled, enacted — don't move at the same pace. That's what makes them so hard to navigate together.
Measured shape updates quarterly. You ship features, the metrics arrive weeks later, the dashboard tells a story about the recent past. Felt shape shifts with every interaction — every user session, every demo where you sense the energy change. Ruled shape lurches at funding events — a new round reshapes possibility space overnight. And imagined shape drifts the most quietly of all, carried by the founder's evolving sense of what this thing is becoming — sometimes so gradually that nobody notices it's moved until it's somewhere new.
When these tempos fall out of sync — and they're always out of sync — the team is navigating with maps drawn at different times. The founder drives toward an imagined shape that the measured shape hasn't validated yet. The board evaluates a measured shape the product has already outgrown. The engineer builds toward a made shape that a market shift just made irrelevant. Everyone looking at the same company, seeing a different moment in its life.
I've felt this dissonance myself — been in the room holding a shape that the data was quietly contradicting, because the story I was telling depended on it.
Drift is the quiet one. A pivot is visible — the team announces a direction, the deck changes, everyone recalibrates. Drift is the shape changing without anyone deciding it should.
Small decisions accumulating. A design partner's use case warping the roadmap by a few degrees each quarter. An integration that turns out to be more popular than the core product. The marketing copy evolving to match what customers actually say instead of what the team intended. One morning you look up and the product is in a different shape than anyone agreed to. It happened in the seams, in the ordinary work of building, and nobody marked the moment because there was no single moment to mark.
The most dangerous version is navigating by a shape the product already left behind. The team still telling the old story. The deck still carrying the old positioning. But the enacted shape — what users are actually doing — has moved. The gap between the story and the reality widens quietly. And when it snaps, everyone is surprised. But the signals were there, in the places nobody was measuring. They always are. Customer language starts shifting — they describe your product in words you didn't give them. The demo that worked last quarter suddenly needs more context. The questions in a meeting shift from "does this work?" to "who is this for?" Each one a signal the shape has already moved.
This is why I keep coming back to shape injection — not because the artifact is right. Usually it isn't. But because it introduces a new temporality into the room. A rough future collapses the distance between where the team is and where they might be. And once you've stood inside a felt shape, even a provisional one, you can't unfeel it. When the market actually moves in that direction — weeks or months later — the team recognizes it. They've already been there, even if only for an afternoon.
The artifact is the part I can do on a Saturday. The part I'm still figuring out is everything around it — when to put it in the room, what tension it needs to create, how much the room can hold. That's not a method. It's a feel for the people, for the founder's relationship to their own vision, for where the company is in its life. I get it wrong a lot.
Nothing holds still. Every shape I've worked with — every positioning, every reframe, every convergence that felt like an answer — was already drifting by the time the room agreed on it. I used to find this unsettling. Now I think it might be the point.
The future you build on a Saturday afternoon won't be the future the founders eventually ship. It'll be something they metabolized, something they made their own. The value was never the artifact. The value was what the artifact made thinkable — the door it opened in a room where all the doors had felt shut.
I used to think the goal was to find the right shape. I'm starting to think it's something quieter: to hold the shapes lightly enough that when they drift — and they will drift — you can feel it happening, and you can follow.