Mission

An agent can write a service, test it, and fix its own bug. Then it stops, because the next step is a browser.

Removing that stop is the whole mission. The rest of this post is why we hold it.

The gap

Engineering changed on my team faster than our infrastructure did. We write purpose-built software for almost every problem we touch now: migration scripts, repair tools for one customer's records, pipelines that run twice and die. Over a hundred one-shot programs in a few months. Before ChatGPT a custom program for one migration cost half a week of attention, so you did it in SQL or you didn't.

Almost none of that software has a home. It runs on laptops against production data, with credentials wherever they happened to land. Not because we don't know better. The only official path to real infrastructure was built for production services, and nobody pushes a 200-line run-twice script through a repo, CI and a Terraform change. The ceremony lost, so the hygiene lost with it.

Writing the software takes twenty minutes. Running it properly takes the rest of the day.

• Write · ~20 min
• Deploy · rest of day

AI collapsed the cost of writing software. The cost of running it barely moved. Every team adopting agents is opening this same gap right now, and the gap is where the risk pools. Production credentials on personal machines. Scripts that changed real data with no record of what ran.

A computer of its own

The agents have the same problem as their output. They run on a laptop that dies when the lid closes, or in somebody else's sandbox that is gone by morning. The thing that writes the software and the software it writes both need somewhere to live.

So the unit we build is a computer. A real virtual machine with root and its own disk, an address that answers over HTTPS and accepts SSH, and keys for the people who can open it. It keeps its files. It sleeps when nobody is using it, and asleep it costs nothing. It wakes when someone opens the URL.

Durable, up in seconds, owned by no model provider. The agents will keep changing. The computer shouldn't have to.

The bar is a Google Doc

Making a Google Doc costs nothing first. You don't create a project or pick a region or attach a billing account. It exists, and it has a URL.

Making a computer should feel like that. One command, and an address comes back.

Sharing should feel like that too, and this is the part hosting usually gets wrong. A Doc is private until you name someone. Then that person can open it, and nobody else can. You don't build a login page or pass a token around in a chat. A running machine deserves the same default.

What we believe

Idle should cost almost nothing. A thousand sleeping computers should cost about what a thousand files cost. If keeping a tool alive costs real money, people kill their tools and rebuild them worse a month later.

Secure should be the default, not the reward for effort. Credentials belong on the machine where the work runs, scoped to that work, not on a laptop. The hygiene has to come free with the path of least resistance, because the last three years proved it won't come any other way.

Small software deserves real infrastructure. Not production ceremony scaled down. A computer shaped for its actual life: born in a chat session, awake for minutes, asleep for weeks.

Where we are

Shed is not open yet. The home page describes the machine as we are building it, and the numbers on the pricing page are not final until it opens. We would rather name the distance than write a mission that quietly implies it closed.

Where this goes

The near picture: you leave your agent running on its own machine, the migration finishes at 2am, and in the morning you open the same session from your phone. The dashboard it wrote is still up.

The far one is bigger. When running software gets as cheap as writing it just did, software stops being rationed. Every recurring annoyance in a company becomes a small program someone made in an afternoon, and all of it needs somewhere to live.

This is for people and agents shipping small software. It is not a replacement for your production estate.

That's the gap Shed closes. If it's yours too, follow along.