Overview

Experimentation and evaluation forms the cornerstone of our platform. This includes Prompts experimentation and testing, end to end testing of your application using Workflows, ability to view and manage all your test run reports, and set up continuous evaluation on logs.

Evaluation can be run on multiple elements - prompts, workflows or directly on datasets with output data.

The below diagrams clarify how the different elements sit together to run evaluations right from experimentation and pre-release to continuous quality check in production.