Agentic AI Meets Search: Real-World Use Cases With...
summary
OpenSearch 3.0 experimental MCP enables agents to call search and observability data as typed tools and the demo wires a Streamlit UI to an MCP connector using GPT-4o-mini (and recommends Claude for best results). Architecture runs OpenSearch nodes with about 33 indices, exposes MCP APIs to register tools and sessions, and the team shrinks prompts from 3.3 million to 60,000 tokens for cost control. Live demos show cluster health, duplicate-index detection, query generation (sometimes incorrect), and recommend read-only guardrails, audit logs, and least-privilege access.
AI-written summary · about
chapters (12)
- 0:00Agentic AI and MCP intro
- 3:36What is agentic AI
- 7:13Why MCP is needed
- 9:22MCP in 90 seconds
- 11:29OpenSearch 3.x MCP features
- 13:13Built-in MCP server
- 16:21MCP client setup
- 18:28Demo setup
- 22:35Live MCP demo
- 31:43Real-world MCP use cases
- 33:31Security best practices
- 35:00Key takeaways
related talks
28mAI Agents in Action: Harnessing MCP and OpenSearch for Modern Use Cases
28mChatOps Using OpenSearch MCP Server
25mKeynote: OpenSearch in the Agentic Era
06mSponsored Keynote: OpenSearch - See Everything: Open Observability for Agentic AI
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
40mAgentic Relevance Tuning With OpenSearch: Faster, Easier, and Higher Quality Search at Scale
42mObserve Smarter, Not Harder: Scaling AI-Powered Observability With O...
43mOpenSearch & AI Driven Cloud Native Platforms: Adaptive & Agentic Microserv...