Sponsored by: KPMG | Building a Governed Synthetic Data Lakehouse for AI Agents

Synthetic data + AI agents is very hot for Databricks right now. The objective of the talk is to propose exploring the idea of synthetic Lakehouse designed specifically for AI agents rather than traditional approach to data collection. The motivation is that teams want to build and evaluate agents over conversations, documents, tickets, and knowledge bases, but real enterprise data is often sensitive or regulated, which blocks experimentation and validation. We will show how synthetic data can be used as a safe, realistic test to develop and evaluate production grade AI agents. We will explore how to design agent-ready synthetic data and why a synthetic Lakehouse is a practical foundation for testing agents even before touching real data.