In this overview video, you'll discover how Deepchecks helps you evaluate and optimize LLM-based agent workflows—from raw input to final output.
The walkthrough covers automatic scoring of individual workflow steps (e.g., search, reasoning, generation), how to define evaluation properties, and how to trace performance back to specific stages.
The video also showcases how you can iterate on evaluation logic using properties, compare runs over time, and pinpoint the root causes of failures or regressions. It’s a comprehensive guide for teams working with complex multi-step LLM systems like agents, looking to boost both quality and observability.
Yaron Friedman
