Agent-powered fleet management system
“Demo provides real-time multi-step vehicle diagnostics with chain-of-thought transparency and MongoDB traceability; live workflow dashboard; resume functionality”
team
Why they built it
To showcase an AI-driven diagnostic system for fleet vehicles using vector search on historical issues and full traceability in MongoDB for actionable advice from complaints and telemetry.
What worked
Multi-step agent workflow, Voyage AI embeddings, Atlas Vector Search, OpenAI recommendations, agent profile management, session tracking, two-column UI for workflow and DB docs
What broke or was painful
Telemetry data simulated via CSV (requires API replacement for production); run-agent-sse API not yet used
The result
Demo provides real-time multi-step vehicle diagnostics with chain-of-thought transparency and MongoDB traceability; live workflow dashboard; resume functionality