Operating Awaken Workforce

Awaken Workforce’s operator promise is simple: you should not need a transcript to know where work is or why it stopped. Business position, scheduling, execution, approvals, and attention are separate typed surfaces.

flowchart LR
  Issue[Issue state] --> Scheduling[Scheduling projection]
  Scheduling --> WorkUnit
  WorkUnit --> Result[Event or output]
  Result --> Transition

Daily loop

1. Create or inspect work in Issues, then switch between list, board, or tree.
2. Open Issue detail and inspect diagnosis/scheduling before dispatching; it names backlog, dependency, Resource, attention, running, ready, or closed state.
3. Read /api/issues/{id}/work-units, then the chosen WorkUnit’s /events and /state.
4. Resolve visible attention on Issue detail; use /api/inbox or /api/tool-approvals when operating across Issues through an integration.
5. Fix the cause behind an attention signal, then mark it resolved.
6. Use audited WorkUnit controls to message, pause, resume, interrupt, redirect, or cancel live work.

Why this earns trust

• Status is a projection of committed facts, not a manually curated label.
• Routing reads declared structured output, never an LLM summary.
• Required inputs and stale approvals fail closed.
• Every privileged execution is fenced by a lease and frozen route snapshot.
• Pack and Resource operations use the same admission and authorization paths as direct API actions.

The product UI is a client of these same contracts. It is the default human surface, while the API supports integration and deeper diagnosis. The UI does not duplicate lifecycle truth: commands and projections remain server-owned.