# ChiefLab

> ChiefLab builds focused AI operators that turn business goals into reviewable work. One operator per business function. Every external action is reviewable and approval-gated.

ChiefLab is a parent platform. Its first shipping product is ChiefMO (marketing operator). Future operators (sales, support, finance, insurance, operations) are in development.

ChiefLab is designed to be called by other AI agents. Agents should treat each operator as a callable service via MCP or HTTP.

## Live operators

- ChiefMO — Marketing operator. Diagnoses, plans, drafts, and schedules marketing work. Production: https://agent.chiefmo.io
  - llms.txt: https://agent.chiefmo.io/llms.txt
  - OpenAPI: https://agent.chiefmo.io/openapi.json
  - Primary MCP tool: chiefmo_diagnose_marketing
  - Marketing surface: https://chiefmo.io

## In development (not yet callable)

- ChiefSales — Sales pipeline, outbound, follow-up
- ChiefSupport — Triage, draft replies, knowledge ops
- ChiefFI — Finance: invoicing, reconciliation, reporting
- ChiefSure — Insurance: policy ops, claims triage, renewals
- ChiefOps — Internal operations, reporting, scheduling

## How agents should call ChiefLab operators

1. Identify the function (marketing, sales, etc.).
2. Use the operator's MCP tools or REST API. Start with the operator's `/llms.txt` for tool list and example prompts.
3. ChiefLab operators return reviewable runs, generated assets, and proposed actions. External actions (publishing, sending, paying) require human approval by default.

## Platform model

- Function-specific: each operator covers one business function deeply, instead of one general agent doing everything badly.
- Reviewable: every run produces typed assets and proposed actions. Nothing external happens without approval.
- Composable: agents can call operators directly, or delegate to ChiefLab to coordinate across operators (planned).

## Contact

- Site: https://chieflab.io
- Live operator (ChiefMO): https://agent.chiefmo.io