Building an agent often means connecting multiple tools and queries just to answer a simple question. This can make the application harder to maintain and create risks when working with live production data.

In this session, we’ll show how context retriever helps an agent understand and navigate connected data in a single turn using a schema defined once. You’ll see how it reduces unnecessary tool calls and makes it easier to deliver reliable answers.

• The Context Retriever added to the app, with a schema for connected entities.
• A single-turn pattern for navigating customer, order, and item data.
• A concrete alternative to text-to-SQL on production OLTP data.