A chief data officer at a health insurance administrator recently told dbt Labs that despite having two data engineers earmarked for AI work, executive support, and budget, his team still can't start. His concern: "Your data is still not at the level where you can put an AI agent on top of that." That gap is the focus of this year's dbt Summit, set for September 15-18 at The Cosmopolitan in Las Vegas.
The event marks a decade since dbt helped create the analytics engineer role, and organizers draw a parallel to today's AI shift. They argue that data consumers are moving from analysts in BI tools to agents running continuously at machine speed, raising stakes on every data decision over the next 18 months.
The main keynote is built on four pillars: leveling up the engine (faster parsing, better scalability, cost control), leveling up for AI (structured, governed context for agents), leveling up with AI (via dbt Wizard and agentic workflows), and leveling up the stack (open, flexible infrastructure). Thursday's Community Keynote features Grace Goheen and Jeremy Cohen, celebrating dbt Core and the engine uniting under one framework.
Product sessions include: dbt State optimization, with Reuben McCreanor showing average compute savings of 30% through metadata checks and auto-deferral; dbt Wizard's three modes (develop, analyze, discover) demonstrated by Ani Venkateshwaran and Brandon Thomson; a Rust-based Apache 2 engine rebuild for dbt Core, presented by Hope Watson with side-by-side parse speed demos; and a session by Zach Mandell arguing the semantic layer is now central to AI strategy, with the dbt MCP server turning governed metrics into agent-ready context.
Other sessions cover open data infrastructure with Apache Iceberg (Jack Lowery and Anna Lee), Apache Ossie's semantic layer portability spec co-led by dbt Labs, Snowflake, Salesforce, BlackRock, and RelationalAI, and Fivetran-dbt integrations for unified lineage and agent context. More sessions will be announced before September.
Source: dbt Labs Blog