Teaching Machines To Debug Machines: Building an AI-Powered RCA...
summary
Built an autonomous RCA agent called Monbot that ingests alerts from Sylva‑Core, calls 80+ tool MCPs (Kubernetes, OpenStack, Grafana/Loki/Prometheus), and stores investigations in a memory layer to correlate past incidents; stack is Python, RabbitMQ, Pocketflow/Langraph orchestration, LLMs (Quen 3, GPOSS 120B, Claude), and Slack notifications. It reduces noisy cubectl describe usage from ~25% to ~5%, recognizes a pattern seen 96 times, and runs human‑in‑the‑loop validation.
AI-written summary · about
chapters (8)
- 0:00AI-powered RCA in Silva
- 1:48Kubernetes cluster challenges
- 4:37Open-source AI tools for Silva
- 9:50Introducing Monbot
- 12:19Building Monbot
- 20:47Memory and infrastructure mapping
- 23:32AI as co-pilot
- 26:46Q&A
related talks
32mGenerative AI for Observability: Automating Root Cause Analysis in Modern IT Systems
07mAccelerating RCA with AI
28mBeyond Dashboards: Architecting AI Agents for Autonomous Observability
27mAI-Powered Root Cause Analysis at Scale: From Theory To Production...
45mAgents, Tools, and MCP, oh my! Next-level AI concepts for developers
29mPairing with the machine: How AI Agents are rewriting the rules of software development
38mBuilding AI responsibly with a GRC strategy
37mRAG, Tooling, MCP, Agents ... and What is Missing?