LLM Evaluation, AI Red Teaming & Observability Platform

Learn how DATUMO Platform’s Evaluation, Red Teaming and Observability modules work together to test and improve AI agents before and after launch.

The quality of an AI service keeps changing, even after you ship it.
Every time you change a prompt, you need to check whether it actually got better, whether it holds up against attacks you didn't anticipate,
and how it answers in front of real users.

DATUMO Platform lets you answer these questions with numbers.
Three pillars, evaluation (Evaluation), attack (Red Teaming), and observation (Observability), feed into one another,
and every new problem you find becomes a criterion for the next evaluation.
The more you evaluate, the closer your criteria get to the real risks of your service, and those criteria become the quality bar your service must meet.

Modules

How the docs are organized

DATUMO Platform Docs is split into Docs and ChangeLog.
Click a name in the left menu (Evaluation, Simulation, and so on) to open its concept document. The Tutorial under it has usage guides you can follow in screen order.

Get started

If you're new, start with the Quickstart.