Building better products on unpredictable models

Freeplay gives teams building products with LLMs a more disciplined way to understand what their models are doing. Instead of managing prompts through code and piecing together model behavior after the fact, teams can test prompts and models, compare outputs, evaluate performance, and monitor production interactions in one place.

I led a redesign across Tests, Prompt Editor, Sessions, filtering, and Datasets to make that workflow easier to inspect and improve.