Beyond Copilots: We're Building a Data Stack Live with AI Agents

​You've seen a hundred demos of AI writing a SQL query. The query was never the hard part. The hard part is everything around it: the ingestion pipeline, the schema wrangling, the dbt models, the metric definitions that keep your dashboards (and your AI) honest.

​So we're building the whole thing live, in under an hour, with one schema doing the traveling. With dltHub, an AI agent generates and deploys an ingestion pipeline for a real REST API. The data lands in MotherDuck's serverless data warehouse. That same schema bootstraps dbt models and a Lightdash semantic layer, so when an AI agent answers questions at the end, it can't make numbers up. Elvis (dltHub), Jacob (MotherDuck), and Oliver (Lightdash) walk through it layer by layer.

​You'll learn how to:

• ​Generate and ship working ingestion pipelines with AI instead of writing glue code

• ​Query data the moment it lands, with no infrastructure to set up

• ​Turn one inferred schema into dbt models and a governed semantic layer

• ​Keep AI agents grounded in real metric definitions