Protege Sharpens Healthcare AI Data Strategy With New Benchmarks, Report and Sales Expansion

Protege is an AI-focused data infrastructure company advancing healthcare applications through curated, compliant datasets and specialized benchmarks. This weekly summary reviews its latest moves in healthcare AI, data strategy, and commercial expansion.

During the week, Protege promoted a new report underscoring that effective healthcare AI hinges on access to controlled, high-value data assets. The company stressed that future models will require both broader datasets and deeper curation to meet clinical, regulatory, and performance demands.

Protege’s messaging highlighted that data ownership, quality, and governance are increasingly central to AI outcomes, particularly in regulated healthcare environments. This framing aligns the firm with segments of the value chain that monetize data platforms, tools, and advisory services for AI developers and providers.

Operationally, Protege is expanding its go-to-market footprint with a Business Development Representative hire for its healthcare team. The role focuses on driving pipeline for AI model development using real-world healthcare data, signaling a tailored outbound engine rather than generic sales coverage.

This commercial build-out is aimed at accelerating customer acquisition in healthcare AI infrastructure, where data and model development are key value drivers. A stronger outbound function could support more predictable revenue scaling and deepen Protege’s position in the health data and AI ecosystem.

On the product side, Protege detailed new medical benchmarks built from “uncontaminated, evaluation-ready” electronic medical record datasets linked to payer-approved bills. By holding out data at the patient level, these benchmarks seek to reduce contamination and inflated performance metrics common in public coding datasets.

The benchmarks emphasize payer-approved claims, aligning evaluation with reimbursement and compliance outcomes for providers and payers. Working with Vals AI, Protege assessed models on ICD code assignment and compliant code-set optimization, using expert coder review to validate results and identify real-world gaps.

Initial findings show models reaching about 88% accuracy on clinical documentation but only 56% on medical coding, highlighting the complexity of structured billing tasks. Protege frames medical coding as an evidence extraction and optimization problem requiring nuanced reasoning around severity, comorbidities, and institutional protocols.

These results reinforce Protege’s view that payer-aligned benchmarks and robust datasets are central infrastructure for healthcare AI, especially in revenue-cycle and administrative workflows. If widely adopted, the benchmarks could support recurring demand for Protege’s data products and evaluation frameworks.

Strategically, CEO Bobby Samuels used an a16z “Raising Health” podcast appearance to argue that data, rather than compute or model design, is now the main bottleneck in AI. He emphasized that unlocking real-world data at scale, while compensating data holders, is core to Protege’s platform and business model.

The podcast also noted that Protege has completed three financings in under two years, led in part by Andreessen Horowitz partners. This venture backing provides capital to expand dataset coverage, refine benchmarks, and build scalable products for highly regulated sectors like healthcare.

Samuels highlighted a partnership-first philosophy in which data owners share in the value derived from their assets, supporting compliant and ethical data monetization. Such structures may help Protege address regulatory and privacy constraints while attracting institutional data partners.

Overall, the week underscored Protege’s focus on closing the AI data gap in healthcare through specialized benchmarks, reinforced data strategy messaging, and expanded commercial capacity, underpinned by sustained investor confidence.