From court to code: Build an agentic RAG assistant with Elasticsearch
Want to see what it really takes to build a smart AI assistant? How about one that can help you make the right fantasy basketball picks?
In this live session, we’ll demonstrate how to instantly activate and ground a high-performance AI agent using the Elastic Agent Builder, and we’ll show how it powers real-world use cases like smarter player picks.
Join JD Armada, developer advocate, for a 20-minute live coding session to learn about:
- Agent Builder Activation: Enabling the Agent Builder in Elastic's serverless environment with no extra infrastructure setup
- Data Grounding: Uploading and indexing an NBA dataset directly into Elastic, and automatically defining the data schema
- ESQL as a Tool: Registering powerful ESQL queries (averages, comparisons) as specialized tools the agent uses for accurate answers
- External Integration: Exposing the agent via MCP so it can be called and reused by external applications and other LLMs (like Claude Desktop)
- Insight Generation: Building a custom agent to answer complex, high-value questions like, “Who’s a better pick for my fantasy team, LeBron or Curry?”
Key Highlights:
- Live demo of an AI assistant that returns real-time, stat-based comparisons
- See how to expose the agent as an MCP service for seamless external integration across your organization
- Observing the agent’s reasoning process with built-in tool call traces and structured responses
Resources:
- Get the dataset for this session
- Search Labs Elastic's blog for developers and data scientists: https://www.elastic.co/search-labs/blog/agentic-rag
If you’re looking to deploy powerful, data-grounded AI agents fast, this session is for you. Jump into the chat, follow along, and see what’s possible with Elastic Agent Builder.
👍 Like, share, and subscribe for more Elasticsearch Coding Sessions!