Monte Carlo data incident troubleshooting agents

What was done

Monte Carlo uses LangGraph for an AI Troubleshooting Agent that starts from a data alert and explores investigation branches such as code changes, timeline events, dependencies, and likely root causes. The graph can spawn subnodes so the system checks multiple hypotheses concurrently instead of following a single manual path. LangSmith helped the team visualize graph runs and iterate on prompts while the production architecture connects existing Monte Carlo APIs with Bedrock and AWS services.

Stack