AIOps: (near) Zero-Touch Production Rollout Fixes
summary
A live proof of concept that plugs an AI Metrics plugin into Argo Rollouts and Argo CD to analyze logs and Prometheus metrics for a 10% canary with a 95% success threshold, detects a null pointer and a memory leak within seconds on GKE, rolls back automatically, and opens PRs or issues for remediation using Gemini LLM or in-cluster models. Built with Quarkus and A2A, it runs fast enough to cut MTTR and continues asynchronous code fixes for human review.
AI-written summary · about
chapters (8)
- 0:00AIOps & progressive delivery
- 4:58GitOps basics
- 7:03Argo Rollouts implementation
- 9:08AI for rollout analysis
- 14:04Demo: AI rollout analysis
- 21:05Demo: Memory leak detection
- 24:43Wrap-up & lessons
- 28:56Q&A
related talks
24mFix Production Rollouts on the Fly With Agentic AIOps
26mBeyond Argo Rollouts: Boosting Developer Experience With Intelligen...
33mSelf-Healing Rollouts: Automating Production Fixes With Agentic AI
30mArgo CD Observability with AIOps - Detect Incident Fast
23mGitOps Meets AIOps: Building Reliable AI Pipelines With Argo
35mFrom Static Rules To Reasoning Platforms: Scaling Intelligent Canary Delivery in 2026
44mAgentic Delivery, Guardrailed: AI in CI/CD & Platform Engineering (Open Source Patterns)
27mIntegrating Agentic AI Into Internal Developer Platforms To Minimize Errors and Downtime