Demo: Mosaic AI Agent Framework and Agent Evaluation

The following demo notebooks introduce you to the Mosaic AI agent development workflow. In these notebooks you:

1. Generate synthetic evaluation data from a document corpus.

   By using synthetic data, you reduce the need for human subject matter experts to label and evaluate results.

2. Create a tool-calling agent with a retriever tool.

3. Evaluate the agent on quality, cost, and latency across several foundational models.

4. Deploy a production-ready agent to a web-based chat app.

You can import these notebooks and run them yourself, or copy code-snippets and ideas for your own use.

10-minute Mosaic AI agent demo

This notebook is designed to quickly get you working with Mosaic AI agents by using a sample document corpus. This standalone notebook is ready to run with no setup or data required.

10-minute Mosaic AI agent demo

Open notebook in new tab Copy link for import

Mosaic AI agent demo - bring your own data

For users that already have a Databricks Vector Search index, use this notebook to go through the same workflow as the 10-minute demo using your own data.

Mosaic AI agent demo - bring your own data

Open notebook in new tab Copy link for import

< Previous: Gen AI tutorial intro

Next: Intro to RAG in AI development >