AI Agent Organization Model
The structure of Hunexia's AI agent organization. The CEO → team-lead (orchestrator) → expert chain, departments, and the Labs standard development chain.
Typically the CEO speaks to the team-lead → the team-lead analyzes and delegates to experts (in parallel where possible) → synthesizes the results. Experts can also be invoked directly.
The CEO sets goals and, without instructing each task, the team-lead (Eden) autonomously runs the organization to match those goals: assigns sub-goals to experts → checks progress → reviews → synthesizes and reports to the CEO. The CEO uses reports to check the flow and correct direction.
operations/ (reports, retrospectives, goal ledger)./report (Operating rhythms).operations/intent-alignment.md (CEO intent notes).platform-engineer, data-analyst, etc.) is used in subagent_type. The generic Explore is limited to preliminary exploration — named invocation is a prerequisite for per-role token visibility (operations/metrics/) to attribute to role names.After each task, every agent accumulates learnings in .claude/org/reviews/<name>.md, and important items are saved via remember.
The structure of Hunexia's AI agent organization. The CEO → team-lead (orchestrator) → expert chain, departments, and the Labs standard development chain.
The HR system for the agent organization. Procedures for hiring, performance review, and career levels, with CEO approval gates. Every organizational change is recorded as a git commit.