Braintrust is a platform to building AI applications, making it more efficient for Large Language Models (LLMs) to become more robust, reliable, and interactive.

Some features include prototyping with different prompts and LLMs in a sandboxed environment, real-time monitoring and performance insights, and data management through their intuitive UI.

Most Large Language Models support tool calling, which allows you to define tools with well-defined input and output types. They are commonly used for two purposes:

  1. To enable models to “call” tools that perform external tasks, and then use those results to produce a final response
  2. To coerce a model into production structured outputs that match a given JSON schema

Braintrust supports both use cases, including using prompts in your code which can be executed directly using TypeScript or Python. Custom functions, a broader concept in Braintrust, enhance the capabilities of LLMs by allowing you to:

  • Add web-browsing capabilities (using tools like Browserbase)
  • Perform complex computations
  • Access external APIs

​
Add Browserbase to your Braintrust

Add Browserbase to Braintrust Tool Calling

Configure Browserbase to add additional web-browsing capabilities to Braintrust tool calling.

Was this page helpful?

Long Running SessionsQuickstart