NYC-based Protege, which prepares and sells real-world datasets like lab results and sports footage for AI training, raised a $25M Series A led by Footwork
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 …
Context & Ripple Effects
Protege’s round extends a line of investment in the data layer beneath AI models: Scale built a managed marketplace for human-reviewed training data, while DatologyAI raised funding to improve training-data curation.
The company is targeting real-world material rather than synthetic inputs, placing it alongside a market where the provenance, preparation and commercial packaging of data are becoming distinct products.
First-order effects
- Protege gains $25 million in Series A capital, led by Footwork, to support its business of preparing and selling real-world datasets for AI training.
- AI developers seeking datasets such as lab results or sports footage gain another specialized supplier alongside established human-data providers such as Scale’s contractor-driven labeling marketplace.
Second-order effects
- The funding raises pressure on data vendors to differentiate through access to particular data types and through preparation quality, not simply by supplying labels.
- Curation and quality-control providers may benefit as buyers seek usable training inputs; Cleanlab’s earlier funding for data-labeling accuracy tools reflects that adjacent demand.
Third-order effects
- If specialized data suppliers continue to attract capital, AI training data may evolve from a generalized labeling service into a more segmented market organized around rights, domain expertise and data preparation.
- That shift could make differentiated data access a more durable competitive input for model builders, while increasing the importance of how real-world material is packaged for commercial use.
The trend: AI investment is moving beyond generic labeling toward specialized, commercially prepared real-world data as a differentiated layer of AI infrastructure.