Zentropi

Build and run content labelers from policies written in plain English, by web app or API.

About Zentropi

Users can create, test, and deploy custom content labelers for both text and images. The service is designed for users who are at least 18 years old.

The service is described as covering use cases such as hate speech detection, PII identification, sentiment analysis, and custom compliance checking. The Services are not tailored to comply with industry-specific regulations such as HIPAA or FISMA.

Drafted from zentropi.ai · not yet confirmed by Zentropi

How it works

Zentropi turns a policy written in plain English into a labeler: a classifier that returns a binary label (match or no match) with a confidence score for a piece of text or an image. Labelers run on CoPE, Zentropi's small open-weight model, which is listed separately.

You can build a labeler in the web builder or through the Build API, test and benchmark it, then deploy it and classify content against it through the Run API. An agent skill, installed with npx skills add zentropi-ai/skills, teaches AI coding agents to write, test, tune and deploy labelers through the same API; it needs a Zentropi account and API key.