Ask Astro: How CAVA Built an Enterprise scale Agentic Data Copilot on Databricks

At CAVA, teams across all business domains rely on governed data in Databricks to drive decisions across the enterprise. This includes datasets used by Investor Relations to prepare earnings materials, support board reporting, and inform market-facing communications.To enable users seamlessly access and act on data across through a single interface, we built Ask Astro, an enterprise-scale agentic data copilot on Databricks that consolidates analytics, institutional knowledge, and operational workflows into one governed conversational experience.Ask Astro is powered by a Multi-Agent Supervisor architecture that intelligently routes user requests to specialized agents and governed tools. The system is deployed in production through databricks apps and as a Slack bot.In this session, we will walk through the end-to-end architecture of Ask Astro, including the Supervisor Agent, MCP setup, internet tool governance, and production hardening practices that enabled safe enterprise scale.