Human ownership
A named operator uses the workspace and decides when work should proceed. Intelligence is cheap; permission is the product.
A practical position on AI
Your LLM can know the fix. A person decides if it runs.
FlowRMM is not a black-box remediator. Your LLM: ChatGPT, Claude, or a local SLM: can stage a named action over MCP. An operator approves it. The trail shows who said yes. Cohort 01 is that control plane: not autonomous remediation, and not a full RMM.
Cohort scope
What we are putting in customers’ hands now.
Approvals where needed
Policy puts a person in front of consequential actions. Low-impact work can auto-run only when you say so.
Evidence for follow-up
Session, control, and action history help the next operator see what happened: and who approved it.
Operator docs