Building Production-Ready AI Agents on Databricks | Agent Bricks Demo

A production AI agent follows one lifecycle on Databricks: index, build, trace, register, evaluate, deploy. Watch Samantha Wise walk through it, from a managed Vector Search index and agent-as-code to MLflow Tracing, Unity Catalog registration, MLflow evaluation with custom guardrails, and deployment to a real-time endpoint with a review app. 🎯 What You'll Learn: • How to build a governed agent with managed Vector Search and the agent-as-code pattern • How MLflow Tracing shows retrieval, tool calls, latency, tokens, and cost per request • How to evaluate agents with built-in and plain-English custom scorers in MLflow • How the same evaluation scores carry from development into production monitoring ⏱️ CHAPTERS: 00:00 - The agent lifecycle: index, build, trace, register, evaluate, deploy 01:57 - Debugging an agent with MLflow Tracing 04:23 - Evaluating relevance, safety, and custom guardrails with MLflow 06:47 - Using the same scores in production 📚 Resources: • Get Started with Generative AI training (free): https://www.databricks.com/training/catalog/get-started-with-databricks-for-generative-ai-3480 • Get Started with Lakebase training (free): https://www.databricks.com/training/catalog/get-started-with-lakebase-5082 • Databricks AI: https://www.databricks.com/product/machine-learning 🤝 Stay connected: Website: https://databricks.com Twitter: https://x.com/databricks/ LinkedIn: https://www.linkedin.com/company/databricks/posts/ Instagram: https://www.instagram.com/databricksinc/?hl=en Facebook: https://www.facebook.com/databricksinc/