Online workshop: Evals 101 for engineers

​Get started with evals and observability using the Braintrust SDK.

​In this workshop, you’ll instrument an agent with the Braintrust SDK and look at traces to see what actually happened across model calls, tool use, and outputs. Then you'll use real failure modes to build datasets, write scoring functions, and iterate on your prompt.

​You should walk away with a practical framework for evaluating AI systems end to end: how to capture the right data, score what matters, and improve quality over time.