I'm Maysam. For about five years I've been telling myself I'd build my own AI project "when I had the right idea." Mocu is the first version I actually open every day — and yes, it's alpha, and it shows.
A desktop workspace where you build your own small AI systems around the work you actually do, instead of adopting a generic assistant and bending your workflow around it. It's text-based, and each project keeps its own context and its own folder, so long-running work doesn't restart from zero every time you open a chat.
Put together, these let you assemble a setup that's genuinely yours: an agent that knows how your project is organized, extensions that do the specific boring thing you always have to do, and an app surface for the part you touch every day.
My read on where this is going seems a bit different from most tools I try. I don't want agents that do my work for me. I want ones that help me do it — that know my project, keep the context, and point at the next useful step instead of pretending to be me.
That's the whole design bet. Mocu is a workbench, not an employee.
It's alpha. There are rough edges, some things break, and some parts will change. I've also been using Mocu to help build Mocu, which is a strange and slightly humbling loop — when it fails at my own task, that's usually the next bug to fix.
I'd rather show up with something imperfect that I use every day than keep waiting for the version that's ready.
Try it and tell me where it breaks. Not "nice project" — the actual friction. If you have an idea for a small system you'd want to build for yourself (an agent, an extension, a workflow), tell me that too; it's the most useful input I can get right now. Issues and PRs are very welcome.
Repo and setup instructions: [https://github.com/kharchangai/Mocu](https://github.com/kharchangai/Mocu)
One question I'm curious about: what's the one thing you'd want an AI tool to remember about your project so you'd stop re-explaining yourself? I keep changing my answer.