From analytics engineer to context engineer: A dbt playbook

🌐 Global-friendly sessions October 14 & 15

AI agents are only as good as the context they read, and most of that context call transcripts, tickets, docs, contract PDFs has been trapped in source systems or untouched due to the complexity of managing those pipelines.

This session makes the case that context engineering is analytics engineering with a new last mile, and shows how to do it inside dbt. We'll cover the best practices dbt Labs uses on its own data: generate context rather than aggregate it, parse and chunk on warehouse-native AI, embed once and re-embed only what changed, layer hybrid search behind a metadata prefilter, and test the shape of nondeterministic output. Then we'll walk through the dbt_context_engineering package, which packages those patterns as composable models and macros so your agents query governed, versioned context straight from the warehouse.

You'll leave with a reference pipeline you can install with dbt deps and a set of golden-query evals to keep it honest.

Meet the speakers

Alex Noonan

Senior Developer Experience Advocate

dbt Labs

Stephen Thibeault

Senior Resident Architect

dbt Labs