Hello, JellyClaw
We're building the Local AI Office: a hierarchy of AI agents — a CEO, department heads, workers — that runs entirely on your own hardware, works through the night, and reports to your Telegram by morning. You define the whole team in one YAML file. That's the product.
Why local-first
Because an AI office should belong to the person who runs it. When agents run on Ollama on your machine, your code and data stay where they are, there's no per-seat bill that scales with your team's ambition, and nothing stops working because a vendor changed a plan. For the tasks that genuinely need a frontier model, we route to the Claude API — your key, your call, task by task. Local by default, cloud by choice.
The roadmap
- CLI MVP — the engine: team YAML, the agent hierarchy, and the five MVP integrations (Telegram, Ollama, Claude API, GitHub, SQLite).
- Open-source launch — the code goes public on GitHub, free to run and modify.
- Dashboard — a visual layer on top of the CLI, offered as premium. The CLI stays free forever.
Where we are today
Being direct: the vision, the architecture, and a dashboard mockup exist; the core engine is in active development. We'd rather ship you this note than a wall of fake logos.
Follow along
Watch the GitHub repo — that's where the MVP lands first. If you have opinions on the planned YAML format or want to shape the roadmap, open an issue. We read all of them.