Conway’s Law says organizations design systems that mirror their own communication structures. It’s one of those observations that sounds obvious until you realize how much it explains. The org chart doesn’t just describe who reports to whom. It describes the shape of everything the company builds.
I’ve been thinking about this since the pitch pause. Not because it proved Conway’s Law. Because the pitch pause was a deliberate intervention against it.
The problem before the pause
Before we ran the pitch pause, we had a communication problem. Not the kind where people don’t talk to each other. The kind where people are talking past each other because they’re operating from fundamentally different mental models.
Some people in the organization were already deep into agentic AI. They’d been building with it, thinking in terms of what it could do, imagining workflows that assumed these tools existed. Other people were skeptical. Not hostile, mostly. Just unconvinced. They’d seen hype cycles before. They had real work to do and didn’t want to chase something that might not matter in six months.
Both positions were reasonable. But they created an asymmetry in how the organization could think about the future. If you apply Conway’s Law, a company where half the people think in AI-native terms and half the people don’t will build products that reflect that split. Some parts will be forward-looking. Some parts will be built the old way because the people designing them haven’t internalized what’s changed. The seams will show.
You can’t design for a world you don’t share. If the people in the room have different assumptions about what’s possible, the design will be a compromise between those assumptions. Not the best version of either.
What the pitch pause actually did
I wrote about the pitch pause before in terms of what people built. 43 repos in two weeks. Internal tools. Sales demos. A happy hour where everyone showed their work. All of that happened and it mattered. As of April, we’re up to 78 repos. The momentum continued well past the pause itself.
But the deeper thing the pitch pause did was reset the communication structure. Not the org chart. The shared mental model.
By giving everyone across the organization dedicated time to explore these tools on their own terms, without pressure, without deliverables, without anyone standing over them saying “this is the future, get on board,” we let people arrive at their own understanding. The sales engineer who was skeptical got to try it on a problem they actually cared about. The support lead who’d never written code got to build something and feel what that was like. The product manager who’d been curious but too busy got two weeks to go deep.
Nobody was told what to think. They were given time and permission to find out for themselves. And most of them came out the other side with a shared vocabulary that didn’t exist before.
The vocabulary matters more than you’d think
Here’s what I mean by shared vocabulary. Before the pitch pause, a product discussion might go like this:
“What if we let users configure their own data pipelines?”
“That’s a six-month project. We’d need a visual builder, validation logic, error handling, a whole new team to maintain it.”
That response isn’t wrong. It’s just operating from a set of assumptions about what building costs. After the pitch pause, the same conversation sounds different:
“What if we let users configure their own data pipelines?”
“What would that look like in an agent-assisted workflow? Could we give them a natural language interface that generates the pipeline config? What if the agent handles the validation?”
The ideas aren’t necessarily better. But the starting point is different. People are reasoning from the same set of possibilities. The skeptic’s mental model has been updated, not because someone convinced them in a meeting, but because they spent two weeks experiencing what’s actually possible.
That’s the Conway’s Law intervention. You’re not changing the org chart. You’re changing the shared understanding that the org chart sits on top of. When everyone in the room has at least a baseline fluency with what these tools can do, the things the organization designs start to reflect that fluency.
From shared vocabulary to AI-native thinking
The pitch pause was in February. It’s April now. The effects are showing up in ways I didn’t fully anticipate.
Product discussions have changed. Not because anyone mandated a new framework. Because the people in those discussions are now asking different questions. “How will someone in the agentic age want to interface with this?” “How should our product complement AI-assisted workflows?” “What does onboarding look like when the user has an agent helping them?”
These questions used to come from one or two people in the room. Now they come from everywhere. From sales, who are seeing how prospects use AI tools in their own work. From customer success, who are hearing about it in support conversations. From marketing, who are thinking about how to position the product for buyers who already assume AI is part of the stack.
That’s the shift Conway’s Law predicts. When the communication structure changes, so does the output. When people across all disciplines think at the same level about what’s possible, the products start reflecting that shared understanding instead of reflecting the gap between the believers and the skeptics.
Respecting the process
Reflecting on the pitch pause, I think worked in part because it didn’t try to force the outcome. If we’d sent everyone to a training session that said “here’s why AI matters and here’s how you should think about it,” we would have gotten compliance, not conviction. People would have nodded along and gone back to thinking the way they thought before.
The pitch pause respected that people need to find their own moment. Some people’s moment was the first day, when they built something in an hour that would have taken a week. Some people’s moment was halfway through, when they realized the tool could help them with something they’d been carrying around for months. Some people’s moment was at the happy hour, watching a colleague from a completely different team show something that changed how they thought about what was possible.
You can’t schedule someone’s moment of understanding. But you can create the conditions where it’s likely to happen. Dedicated time. No pressure. Real problems to work on. Permission to explore without it interfering with existing commitments and deadlines. That’s what the pitch pause provided. Not a destination, but the space to travel.
The products are changing
This is where it gets interesting. We’re now in the phase where the shared understanding is becoming visible in the products themselves. Design reviews include questions about agent interaction patterns. Roadmap discussions account for capabilities that didn’t exist six months ago. Technical architecture decisions factor in how AI agents will consume and extend our APIs.
None of this would have happened if the pitch pause had only reached engineering. Conway’s Law doesn’t care about one team’s understanding. It cares about the communication structure of the whole organization. A product team that thinks in AI-native terms but works with a sales team that doesn’t will build something that looks AI-native on the surface but doesn’t account for how customers actually buy, deploy, and adopt.
The pitch pause reached across orgs. And now the conversations that shape our products include AI-native thinking from every direction, not just the technical one. Sales asks how agents will change the buying process. Marketing asks how to reach people who are building with these tools. Customer success asks what support looks like when the end user has an AI copilot. Product asks what the interface should be for the agentic age.
These are all the right questions. And the reason they’re all being asked at the same time, by people across every function, is that we took two weeks to let everyone build a shared understanding of what’s possible.
What this doesn't mean
It's important to clarify that I'm not implying we've deluded ourselves into thinking we can use these tools to ship some factor of X faster (though we hope we can) or to magically create more quality features without direct human involvement. We're not there yet (remind me to revisit this in 12 months). But it does mean that we're facing in a direction where allowing those realities to manifest, and having real conversations about how our processes and tools change in a world where some of this seems like it IS increasingly possible.
When we are internally unified in the new art of the possible, we have a lot more productive conversations about the things we need to be thinking about solving and the new problem spaces being placed in front of us, versus spending cycles debating "are these tools and technologies actually worth pursuing?"
Conway’s Law in reverse
Most people encounter Conway’s Law as a diagnostic. “Oh, that’s why our API is structured this way, it mirrors the team boundaries.” It’s usually invoked to explain why something is messy.
But it works the other way too. If you want to change what your organization builds, change how your organization communicates. Not the meetings. Not the Slack channels. The shared mental model that underlies all of it.
The pitch pause was a Conway’s Law intervention disguised as a learning event. Two weeks of safe exploration that didn’t just teach people tools. It gave the entire organization a common language for thinking about the future. And now the things we’re building sound like that common language.
You can’t design for a world you don’t share. So we made sure everyone shared it first.
