The first one's out

A few months ago I wrote about a micro-vc that isn't one. A small group of friends pooling experience, not money. Each of us building small companies on our own, but not alone.

The first product from that experiment is live now. It's called MarkMesh.

I don't want to spend much time on what it does. The MarkMesh blog does that better. I want to talk about why this was the first one, and why it felt obvious.

The shape we know

Tobias and I have spent most of our lives moving data between registries. Domains, mostly. A domain registry is a regulated, jurisdiction-bound source of truth. It updates on its own schedule, in its own format, with its own opinions about what things mean. There are hundreds of them. Keeping them in sync, making one clean view out of hundreds of feeds that are each a little broken in their own way, that's the work we've done for twenty years.

Trademark offices are the same problem with different names.

So when we looked at what to build first, we didn't chase the biggest market or the trendiest wave. We went looking for the problem we were already overqualified for. That was a deliberate choice, and I want to be honest about why.

Why we like the mess

Here's the thing about regulated data that took me a long time to learn. The messier and more fragmented it gets, the more a clean, working view of it is worth. Every office that changes its portal, breaks an API, publishes late, or quietly disagrees with its neighbour about what a status means. Every one of those makes the problem harder, and a working solution more rare, and more valuable.

Most people look at that fragmentation and see cost. We've learned to see it as the thing that protects us. You don't earn that by being clever. You earn it by being the one still standing, still in sync, when the sources around you keep misbehaving.

That only works if you built for it on purpose.

We built the boring part

We didn't rent the infrastructure. We built it. Every adapter, every normalisation, every queue, every reconciliation. Ours, running on machines we control. I'll write more about why another time. The short version: when your whole product depends on fetching from forty sources that break in forty different ways, you don't want a layer between you and the metal that you can't fix at two in the morning.

Owning the unglamorous hard part is unfashionable now, in an age of managed everything. It's also the only way we know to build something that gets more reliable the more the world throws at it.

We do less, on purpose

MarkMesh doesn't file trademarks. It doesn't renew them for you or pay your official fees. That was deliberate. The moment we became a filing agent we'd inherit a thousand jurisdiction-specific ways to fail, and a liability profile that would make us cautious and slow. So we do one thing well. We watch the offices, keep your records in sync, and tell you what changed.

Taking the filing out didn't make the product smaller. It made the part we kept actually good. I think about that a lot. The default move, when something feels complicated, is to add a feature. The better move is usually to remove things, not pile more on.

Why not just a startup

MarkMesh isn't a venture-funded company. It's the first of what we hope will be a few small, independent businesses, each one standing on its own. Tobias and I, and the network around us. No rounds to raise. No team to feed. No clock ticking down.

That isn't a compromise. It's the point. When you don't owe anyone a timeline, you can afford to build the boring part. You can take problems off the table instead of piling them on. You can let the messiness of the world do the work of making your data more valuable. You play a long game because nobody is charging you rent on the clock.

MarkMesh is out. There will be more. The experiment continues.