W1 is for
the dreamers.
The ones with the whole thing in their head and no team, no budget, and no one coming to help. Describe it - W1 builds it, checks it, and shows you every step it took. An AI project agent, built in Africa.
Proudly African
Why we built it.
The harness shouldn’t be
the thing you can’t afford.
Most agents burn their budget rediscovering what they already knew - rereading the same files, resending the same context, thinking hard about nothing. W1 is engineered around useful work per call, so the same money does more work. You shouldn’t be locked out of a serious harness because someone else can afford a model you can’t.
- 01Install on Windows, macOS, Linux, or VS Code - two minutes.
- 02Open a project folder and say what you need.
- 03Watch the work happen, and keep what it ships.
FAQ
What can it actually do?
Real project work, not just chat. It reads an existing codebase and ships changes into it, runs the tests and tells you what passed, turns messy notes into a document, a deck and a spreadsheet, pulls research together with the sources attached, and works across all of it inside one project. If a task takes you an afternoon of tabbing between things, that is the shape of task it is built for.
Do I need to know how to code?
No. You describe the outcome in plain language and W1 does the work in front of you. Knowing what you want is the requirement. But it is not a magic box either - because every step stays visible, you end up understanding more of what happened than you would from a tool that just hands you an answer.
What happens when it gets something wrong?
You see it, which is the whole point. W1 runs the checks that matter for the task and reports what actually failed instead of finishing with a confident summary. Every file it touched is listed, and you can undo the work it did without reaching for git.
Does my project leave my machine?
Your project stays local. W1 sends only the context needed for the request, never your whole codebase, and it will never ask you for a provider API key. Run traces are written to a folder inside your own project.
How is this different from the other agents?
Two things. It works across documents, data and research as well as code, in one project rather than four tools. And it is engineered around useful work per call - reusing stable context instead of rediscovering it every turn - because we think a serious harness should not be something you are priced out of.
What does it cost?
Nothing during beta, and there is no card to add. We will announce pricing before the paid service starts, and we will tell you before anything changes for you.
Who is building this?
A small team in Africa, shipping releases most weeks. Bug reports from beta users have turned into fixes in the same week. If you mail us, a person who works on the product reads it - see contact.
Open a project. Describe the outcome.
Free during beta on macOS, Windows, and VS Code. Sign-in keeps usage fair and your sessions secure.