Aktum beta

Agents

Agents do the work. People make the calls.

Talk a feature through with three agents, then hand the signed spec to a coding agent that runs on your own machines.

01 · Chat

Talk it through before anyone writes it down.

The notes agent keeps a scratchpad of what was agreed. The repository agent answers questions about the code. The review agent says what is missing. When the talk is done, start a room from it.

02 · Runs

The coding agent runs on your sidecar.

Claude Code today, as a local process or in a Docker container, in a separate working copy for each room. Watch each phase, each file and what it has cost. Stop it, with a reason, or steer it. A question the spec does not answer stops the phase and goes to its owner.

03 · Agent log

Every run, explained afterwards.

The log says what each run planned, what it committed and where it stopped, and lists what the AI was not allowed to do.

See it on your own repository.

Request early access

A person reads every request and replies.