More than 80% of AI projects fail because customers aren't ready to deploy AI where it could cause reputational or financial harm. This session changes that. We'll start by introducing Agent Bricks, a revolutionary approach that uses novel research techniques to automatically generate domain-specific synthetic data and task-aware benchmarks. Based on these benchmarks, it automatically optimizes for cost and quality, eliminating the tedious trial-and-error of current approaches to deliver production-level accuracy from the start.

Then we'll dive deeper, showing you how to build custom AI agents from scratch using Databricks' Mosaic AI platform. You'll learn to design multi-agent systems that combine retrieval, reasoning, and generation to answer complex questions grounded in your enterprise data. We'll cover experiment tracking, performance optimization, and building reliable, governed workflows at scale.
Whether you're starting your AI journey or scaling existing efforts, this session provides both the out-of-the-box solutions and foundational knowledge to build next-generation GenAI solutions ready for production deployment.