Breaking
Preventive Medicine

AI Startup Knit Health Aims to Fix Healthcare’s Matching Problem

By Elise Dubois 3 min read
AI Startup Knit Health Aims to Fix Healthcare's Matching Problem - ai healthcare
Knit Health CEO Jonathan Kolstad spoke at the HLTH conference about AI in healthcare.

U.S. healthcare is advanced in treatments but lags in infrastructure. Patients face long waits and high costs for basic care. Jonathan Kolstad, CEO of Knit Health, calls this a matching problem. The issue isn’t the system’s capacity or capabilities but its inability to get the right patient to the right care at the right time, Kolstad explained at the HLTH conference.

Knit Health is developing an AI model called Med-BEAM to address this. Trained on 140 million patient records, it aims to predict patient needs and guide their medical journey. The startup recently secured $35 million in Series A funding to accelerate its commercial deployment, launched out of UC Berkeley last year.

From self-driving cars to healthcare navigation

This approach differs from traditional AI in healthcare. Instead of focusing on diagnosis or treatment, Med-BEAM acts as a world model of the entire healthcare ecosystem. It’s designed to work alongside large language models (LLMs) to enhance efficiency and safety, making it more than the sum of its parts.

Scaling data to clone clinical expertise

Knit Health’s edge lies in its massive dataset. With access to the largest electronic medical record dataset in the world, the company trains its model on billions of clinical decisions. This scale allows Med-BEAM to learn the semantic structure of medicine—essentially cloning the expertise of clinicians. The model captures the collective knowledge of doctors, which is often unwritten and embedded in their experience.

Unlike competitors like Mayo Clinic or Epic’s Cosmos suite, Knit Health emphasizes customization. Med-BEAM can adapt to the specific patterns of individual health systems, offering tailored insights rather than one-size-fits-all solutions. For instance, it can identify why Hospital A outperforms Hospital B in triaging, enabling Hospital B to adopt best practices.

Knit Health is initially targeting hospitals rather than payers. Kolstad explains that building trust with doctors is key for clinical AI adoption. However, the company sees alignment with payers in reducing costs and keeping patients in-network. For example, long wait times often drive patients out of network, a problem Med-BEAM can address by optimizing referrals.

Deployment is slated for early 2027, with the first systems going live in February or March. Interest is high, particularly among safety-net hospitals facing resource constraints. One design partner identified 97 potential applications after starting with just one use case. The model’s ability to match supply and demand has created an “Aha!” moment for many health systems.

Specializing in Health System Patterns

Med-BEAM’s ability to specialize is a key differentiator. For example, in rheumatology, some systems are booking out 12 months, and many patients who wait that long don’t even have rheumatologic disease. Med-BEAM can optimize this by ensuring patients are directed to the right specialty sooner.

Elise Dubois

Leave a Reply

Your email address will not be published. Required fields are marked *