Knit Health, a UC Berkeley spin-out that builds AI for routing patients through the healthcare system, announced on 16 November 2026 that it had raised $35m in a Series A round led by NEA.
Frist Cressey Ventures, Town Hall Ventures and SignalFire took part, and earlier backers Uncork Capital, Moxxie Ventures, The Aligned Fund and Coalition Operators also invested again, chief executive Jonathan Kolstad wrote in a company post. Kolstad first disclosed the round to Axios Pro, according to TNW.
The round brings Knit's total funding to $46.6m, according to TNW. The company did not disclose a valuation.
“We envision a world in which clinical intelligence is no longer scarce, and every patient can be connected to the care best suited to their needs, at the moment they need it.”Jonathan Kolstad, Co-founder and CEO, Knit Health
What Knit Health builds
Knit calls its main product, Med-BEAM, a world model for healthcare. Rather than learning from medical text, as large language models do, it learns from the record of what actually happens to patients: how conditions develop, what clinicians order, refer and diagnose, and what follows each decision. TNW reported that it draws on billions of healthcare decisions and hundreds of millions of patient records tracked over time.
The company says it trains only on data it has documented rights to use, and that the data is de-identified before Knit receives it. The model checks its predictions against real outcomes, so it can signal when there is too little evidence to be confident.
Knit compares the approach to the world models used for self-driving cars and robots. Kolstad wrote that language models and Knit's model are meant to work together, with chatbots and AI agents acting as the interface and Med-BEAM supplying an understanding of each patient's likely path through care.
The company's performance claims
In its own benchmarks, published alongside the funding, Knit tested Med-BEAM against models from frontier AI labs on predicting the clinical actions taken for hospitalised patients over the following 24 hours. The company says its model was up to about four times more accurate, 100 times faster and 250 times cheaper. These results come from Knit and have not been independently reviewed.
Kolstad said the speed and cost matter as much as accuracy, because the model would need to run for every patient wherever decisions are made. He gave the example of a patient reaching the right specialist after one visit instead of three.
What the money is for
Knit said the Series A will pay for larger scientific and engineering teams and for bringing Med-BEAM into more clinical settings. Its first uses are patient navigation, specialist routing, scheduling and hospital operations. NEA partner Blake Wu also listed care management, clinical trial recruitment, hospital-at-home programmes and predicting how patients respond to treatment as possible uses, according to TNW.
Kolstad said Knit does not plan to own every application and wants health systems, technology companies and developers to build on its model.
The funding so far
Knit was founded in 2025 by Berkeley scientists. Its co-founders are Kolstad, chief scientist Maya Petersen, co-CTO Jonas Knecht and chief operating officer Ted Robertson, according to TNW.
The company came out of stealth in May 2026 with an $11.6m seed round co-led by Uncork Capital and Frist Cressey Ventures, with Moxxie Ventures backing an earlier pre-seed and Coalition Operators participating, according to HIT Consultant. At that stage its model was trained on Truveta health record data covering more than 130 million patients across 30 US health systems.
Company website: knit.health
