Unitary specializes in video moderation, incorporating visual, audio, and textual signals into a single algorithm. It processes over 25,000 frames of video per second with human-level accuracy via an API.

In order to achieve high accuracy, Unitary uses multimodal algorithms that take into account different elements such as text, images, captions, and meta descriptions simultaneously for analysis.

The company has productized its core technology for two use-cases: (1) brand safety (i.e. classifying video/image content according to the GARM framework or custom policies) and (2) platform safety.