Four Steps to Building an AI Research Agent, Starting with the Map with Andrew Odewahn

An AI research agent needs to know where to look and how your information systems fit together. In this clip from _Live with Tim O’Reilly,_ O’Reilly CTO Andrew Odewahn outlines four steps for building one: map your sources as a hierarchy, connect them to an LLM through MCP, write a skill that guides the agent, and add Expert Intelligence. Andrew found that listing systems such as Productboard, GitHub, and Jira wasn’t enough. The agent began producing useful results only when he explained how those systems relate and the order in which it should search them. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia