Imagine hiring a small team overnight — a strategist, a developer, a marketing person, a systems person — and coming back in the morning to find they'd actually gotten things done. That's the feeling wanman is going for.
It's a piece of software you run on your own computer that coordinates multiple AI agents, each given a specific role, each aware of what the others are doing. They pass work between themselves, share notes, and produce results — without a human in the loop for every step.
What makes this different from most AI tools is where it runs: on your machine, not on someone else's servers. Your files, your data, your conversation — none of it leaves your building unless you decide otherwise. There's also a hosted version at wanman.ai if you'd rather not deal with the setup.
For a business owner, this is worth watching. Tasks that today require you to brief an AI, wait, read the result, then brief it again — that whole back-and-forth could eventually just... run. You describe the goal, and a small AI team figures out the steps.
We're still early. But this is the shape of where things are heading.
AI agent — An AI that doesn't just answer questions but takes actions, makes decisions, and works toward a goal over time. Less like a calculator, more like an intern with initiative.
Multi-agent system — Several AI agents working together, each with a different job, coordinating like a small team rather than one person doing everything.
Local / self-hosted — The software runs on your own computer or server. Nothing passes through a third-party company's systems.
Open-source — The code is public. Anyone can read it, check it, or improve it. A mark of transparency in a world full of black boxes.
If you're curious, the project is at https://github.com/chekusu/wanman — and the sandbox you can try without any setup is at https://wanman.ai.