databricks-logo

mosaic-ai-agent-demo

(Python)

Create, evaluate, and deploy an AI agent

This notebook walks you through building, evaluating, and deploying an AI agent that combines retrieval and tool usage. You'll work with a pre-chunked subset of Databricks documentation as your dataset.

Supporting documentation

For a comprehensive understanding of this notebook's contents, including the rationale behind the code and the challenges it addresses, see the accompanying Databricks documentation page. (AWS | Azure)

Install Dependencies

Create an agent and tools

Connect to an LLM

Load Dataset

Create Keyword Extraction Unity Catalog Tool

Inspect and test the tool

Create a client-side keyword search retriever

Define create agent factory function

Test the agent

Wrapping it in a ChatAgent

Test ChatAgent

Using Model Config

Write file for models from code

15

    16

    17

    18

    19

    20

    21

    ;