Prodigy is a downloadable annotation tool and Python library for developers building and improving AI and machine learning systems. It supports data creation, review, and model training across natural language processing, computer vision, and audio and video tasks, including named entity recognition, text classification, information extraction, image detection and segmentation, and evaluation. Prodigy also offers workflows for language model training, prompt development, and custom annotation projects.

Designed for customization, the tool combines a web application, command-line commands, and pre-built workflows with the ability to create custom data feeds, interfaces, and Python-based recipes. Models can assist with annotation, helping teams focus on valuable decisions and work more efficiently. Prodigy integrates closely with spaCy and can also be used with other libraries and tools.

The software runs on the user’s own machines, keeping data under their control and supporting offline or air-gapped use. Customers can adapt it to their infrastructure and retain ownership of the models they create, without vendor lock-in.