Companies like Scale AI and Surge have proven there’s a market for human-labeled data, like professionals’ answers to complex math or law questions, that AI research labs can use to train their models.

Now, a young startup is tapping a similar opportunity, linking large model makers and AI startups with organizations that naturally generate valuable data in the course of their work, such as medical labs, production studios and soccer leagues. The startup works with these types of data suppliers to organize and clean up their data and then sell access to the resulting comprehensive data set.

Meet Protege, a one-year-old New York-based startup, that just raised $25 million in a funding round led by early-stage venture firm Footwork to expand these match-making services, according to CEO Bobby Samuels. The 20-person startup gathers groups of real world data—say lab results or sports video clips—and sells access to AI startups that need the information to train their technology.

Earlier this summer, for instance, Protege signed a deal to access footage put together by Polish soccer league Ekstraklasa. The dataset is helping AI labs train video generation models on natural movement, such as players’ kicking and turning. Protege also helps healthcare companies sell their notes or medical images to buyers, after scrubbing the data of identifying information and making sure it is compliant with medical privacy laws. It takes between a 25% to 30% cut, on average, of the resulting licensing deal.