Overview

An AI app is a self-contained workspace that holds everything needed to technically evaluate one AI use case โ€” its models, datasets, and evaluations all live inside it and stay isolated from other AI apps.

  • AI apps are self-contained containers, i.e. all entities (e.g. models, datasets) and evaluations live inside one AI app and do not impact entities in other AI apps.
  • In the CLI, AI apps behave similarly to Git branches - an AI app needs to be created and switched to before making any changes or running new evaluations. An AI app is identified by its key (cf. branch name).
  • An AI app should contain all artifacts and entities required to execute a technical evaluations of a given AI agent/system.

What you can manage

Models
Integrate inference endpoints from external providers or your own infrastructure.
Datasets
Bring your own data as datasets and generate variations with dataset generators.
Secrets & access
Manage secrets, users, and multi-tenancy across your control plane.

Learn by example

My First AI App
Define a complete AI app for the airline chatbot use case.