Chris Krysinski

After migrating from Snowflake, Addepar ventured deep into Databricks' native AI/BI suite. This talk shares our practitioner's roadmap for accelerating "speed to value" by building analytics directly where data lives. We explore why we chose AI/BI over external tools and how we serve as early adopters pushing the platform's limits. Key takeaways: - The AI-Assisted Analyst: Leveraging Databricks Assistant to chat through notebook discovery, bypass syntax struggles, and cure "analyst's block" with AI-suggested Dashboard visuals. - Metadata as Code: Why rich Unity Catalog metadata is the critical UI for accurate GenAI. - Metric Views > Tables: Defining semantics once for both Dashboards and Genie to ensure consistent answers, rather than relying on static tables or views. - Genie & Agents: Tips for "teaching" Genie spaces and early experiments using Agent Bricks to orchestrate structured analytics alongside unstructured knowledge agents.