Scout
Research and discovery
Status · Reading the brief
People + agents
People decide. Agents do the legwork. The four below are ours. On a client engagement they work inside the tools we agree, and nowhere else.
Research and discovery
Status · Reading the brief
Build, data and evals
Status · Running eval 14/20
Deployment and monitoring
Status · Watching production
Cost and usage reporting
Status · Reconciling spend
Status lines show the kind of work each agent reports. They are not a live feed of a client system.
1. A person sets the task and the bounds: which tools, which data, what "done" means.
2. The agent works inside those tools, and only those tools.
3. The action log is reviewable. Who asked, what ran, what came back.
4. A person accepts, edits, or rejects the result. Nothing important ships on an agent's say-so.
No agent ships a change to production alone.
Credentials stay with the client. Agents use scoped tokens, and those tokens expire.
Logs are kept for the engagement and handed over with the work.
We don't name an agent after someone else's product. The roster is Scout, Forge, Relay and Ledger until we have a reason to change it.
The next step
Tell us where you are on the journey. In 30 minutes we'll map the next step, from the first idea to scaling what you have.