Protege

AI companies need massive volumes of high-quality, licensed real-world training data, but acquiring it compliantly is fragmented and difficult.

Protege aggregates data from trusted providers via licensing agreements while providing technical expertise for curating and optimizing datasets for AI training and evaluation.

As frontier models scale, demand for diverse training data outpaces supply. Regulatory scrutiny on data provenance makes licensed data essential.

Works with the majority of the Magnificent Seven. 100+ data providers, 300K+ hours of video, 500K+ hours of audio, billions of clinical notes. Raised $65M total.

Bobby Samuels (CEO), Travis May (Chairman, co-founder of LiveRamp and Datavant), Richard Ho (CTO).