Workshop: Build your eval foundations

Every great agent improves through a continuous loop: observe behavior, evaluate performance, and iterate with confidence.

In this hands-on workshop, you'll build that loop from scratch.

Working step by step alongside the Braintrust team, you'll instrument a real agent, inspect traces, create your first evals and scorers, and learn how to turn agent behaviour into measurable improvements.

You'll leave with a practical workflow you can use to:

Observe what your agent is doing

Build evals that measure quality

Continuously improve agent performance

Whether you're just getting started with evals or looking to strengthen your foundations, this session is designed to give you the building blocks you'll use every time you build and ship an agent.

Bring your laptop and stay for the rooftop drinks.