Dev.to
7/27/2026

Multi-agent AI system for autonomous production incident investigation
Original: I Built an AI Team That Investigates Production Incidents Better Than Most Engineers
Short summary
TattvaAI is an AI-powered incident investigation platform built for the SigNoz hackathon that uses six specialized AI agents orchestrated via LangGraph to autonomously investigate production incidents. Each agent covers one observability domain — traces, logs, metrics, dependencies, alerts, and production changes — gathering evidence that a reasoning engine combines to identify probable root causes. The system aims to eliminate repetitive investigative work so engineers can focus on decisions and faster restoration.
- •Six specialized AI agents each handle one observability domain: traces, logs, metrics, dependencies, alerts, and production changes
- •LangGraph orchestrates agents to autonomously correlate evidence and identify root causes
- •Built for SigNoz hackathon; targets the investigation bottleneck that consumes most incident response time
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



