Building Self-Healing Infrastructure: Agentic AI & Vector Search in OpenSearch
summary
Claims OpenSearch becomes the operational “brain” by ingesting vectorized logs and runbooks, using retrieval-augmented generation to correlate new errors with historical incidents and propose fixes. Architecture couples OpenSearch, Kafka, an LM gateway (example Nova Pro), embedding models, and an agent core that can call Kubernetes/ECS APIs to restart tasks; demo simulates an outage, vector-search matches a runbook and the agent generates and executes a restart successfully. Notes real-world scale (7M customers), a 10-minute model outage lesson, and ~70–80% of early production AI apps fail without controls and human-in-the-loop.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 3:13Observability to self-healing
- 8:29Self-healing architecture
- 18:06Live demo
- 27:49AI adoption steps
- 29:01Q&A
related talks
43mOpenSearch & AI Driven Cloud Native Platforms: Adaptive & Agentic Microserv...
28mUse OpenSearch as Your Vector Search Database To Optimize GenAI Accuracy!
42mObserve Smarter, Not Harder: Scaling AI-Powered Observability With O...
06mSponsored Keynote: OpenSearch - See Everything: Open Observability for Agentic AI
46mDelivering the next wave of innovation with OpenSearch
38mA Visual AI Search Flow Builder for OpenSearch: Build GenAI and Vector...
36mBuilding Next-gen Search With OpenSearch
40mBuilding Intelligent Search Systems With Agentic RAG and OpenSearch