Every run, step by step.
Hand work to an AI agent and you do not have to sit and watch it. It runs in the background, and afterwards you can read back exactly what it did: every model turn, every tool it reached for, and how long each one took. When it gets to something permanent it stops there and waits for a human, rather than deciding on your behalf.
A name that outlives its keys
An AI agent keeps one identity no matter how many times its keys are replaced, with the human it acts for recorded beside it, and a switch that is checked every time it asks for anything. Turn the switch off and the next request is refused. Nothing is undone behind you, and nothing is left ambiguous.
An AI agent you write starts unapproved
An AI agent you write is a model, a set of instructions, a list of what it may use, and a version number. It is listed from the start and approved by a human, never by itself. A new version has to score at least as well as the one already running, judged at the cautious end of the range, or a named human overrides it in writing, on the record.
A run that stops is not a run that failed
Work is queued and done in the background. When a run reaches something permanent it parks, keeps everything it had done so far, and picks up from the same place once a human decides. A no is not a dead end either: the AI agent is told, and can take another route.
Instructions, per project, then shared once approved
An AI agent works from an instruction file somebody wrote. It can be attached to one project and taken off again, and it reaches the shared library only through an approval, with a fresh security scan on the reviewer's card.
Playbooks
A playbook is the rulebook a project runs under: which instructions are required, which checks have to pass, who approves what, who signs it off, and how permanent an action is allowed to be before a human has to be asked. Apply one to a project and the rules travel with the work, enforced rather than displayed.
Register your first AI agent with us on a call.
Also in AI management system
- SandboxA machine of its own per session, and your keys stay out of it.Read the page
- Command lineThe terminal your engineers already use, on your rules.Read the page
- MCP serverConnect any AI agent, and the rules come with it.Read the page
- Client SDK and APIOne typed client for the whole platform, given to your team.Read the page
- Apprenticeship programmeA registered programme where you advance by what you can do.Read the page