Online workshop: Trust the agents you build on your data
Registration
About Event
Agents ship fast and fail quietly. An agent writing SQL against your data can run cleanly and still return the wrong answer.
In this session presented by Braintrust and MotherDuck, you'll see how to connect an agent to your data and give it the right context for accurate answers, then how to evaluate it in production. You'll see the full loop live: build the agent, turn real usage into an eval, set a quality bar, and catch regressions before they ship.
What you'll learn:
How to build an agent that answers questions over your own data
How to give that agent the context it needs to return accurate answers
How to turn real usage into an eval dataset
How to set a quality bar and catch regressions before release