Multi-agent operations orchestrator — the capstone of a 5-agent agentic AI platform.
Active development begins Week 34 (Phase 5). This repo is the capstone of the portfolio — full platform launch, portfolio site, and NVIDIA NCP-AAI exam at completion.
ops-orchestrator is the top-level router for the entire agent platform. It classifies incoming requests, dispatches them to the appropriate specialist agent (or chains of agents), manages end-to-end workflow state, and returns a unified response. The portfolio site also lives here.
An intent classifier routes requests to one of four specialist agents. Multi-step workflows (e.g., new request → intake → procedure lookup → work order creation) are managed as sequential chains with shared state. A supervisor layer handles error recovery and retries.
| Metric | Target | Method |
|---|---|---|
| E2E Success Rate | ≥85% | 20 end-to-end scenarios |
| Routing Accuracy | ≥95% | 50 varied requests |
| Full Pipeline Latency | <30s | Automated timing |
This orchestrator connects all five agents. At Phase 5 completion, agent-core goes public and the full system is visible E2E.
| Agent | Pattern | Status |
|---|---|---|
| knowledge-navigator | RAG Agent | 🔄 Active |
| intake-agent | Structured Output | ⏳ Phase 2 |
| compliance-checker | Multi-Agent Pipeline | ⏳ Phase 3 |
| workload-scheduler | Planner-Executor | ⏳ Phase 4 |
| ops-orchestrator | Orchestrator | ⏳ Phase 5 |
At go-live: connect each specialist agent to your organization's data sources via the DAL config. The routing logic, workflow chains, and orchestration layer are unchanged.