Knit Health raises $35M from NEA to model how patients move through care
The Series A follows an $11.6 million raise reported in May. Knit says its Med-BEAM model learns from what clinicians actually do, and it plans to start with patient navigation, specialist routing, scheduling and hospital operations.

- Amount
- $35M
- Type
- Series A
- Company
- Knit Health
- Lead investor
- NEA
- Sector
- Health AI
- Announced
- Nov. 16, 2026
Knit Health is betting that the hard part of health care is not knowing medicine but knowing which patient should get which care, and when. The company said Nov. 16 that it raised a $35 million Series A led by NEA to build an AI model aimed at that question.
Frist Cressey Ventures, Town Hall Ventures and SignalFire joined the round, and earlier backers Uncork, Moxxie, the Aligned Fund and Coalition Operators invested again, according to a post by Jonathan Kolstad, Knit's co-founder and chief executive. Knit's newsroom lists a May 12 Endpoints News report that the company had raised $11.6 million. Knit did not disclose a valuation.
Why it matters: Knit is pitching an alternative to large language models, which Kolstad credits with fueling much of the recent enthusiasm for AI: a model trained on the paths patients take through care rather than on text. Its first targets are operational jobs, such as routing patients to specialists and scheduling.
What Knit is building
Kolstad describes the company's model, Med-BEAM, as a "world model" of health care. He said it has learned the routes patients take through the system, including how their conditions change, what clinicians do, what follows each treatment choice and how all of that differs from one patient and setting to another. He compared the approach to the models used to guide self-driving cars and robots, and said Knit sees such a model as complementary to language models, which can serve as the interface for patients, clinicians and AI agents.
"Knowing what the literature says about a condition is not the same as knowing what this patient needs next," Kolstad wrote.
The company says that on tests of forecasting actual clinical events and care decisions, Med-BEAM scores about four times better than leading language models, and does so 100 times faster at 250 times lower cost. Those figures are Knit's own; the post did not name the language models it compared against. Maya Petersen, a co-founder and Knit's chief scientist, published the company's early benchmark results the same day.
Where it starts
Knit plans to begin with patient navigation, specialist routing, scheduling and hospital operations, Kolstad wrote. He said the company does not plan to own every application and would rather let health systems, tech companies and outside developers build on its model. One example he gave: getting a patient to the right specialist on the first visit rather than the third.
The Series A will pay to grow Knit's scientific and engineering teams and to bring its technology into more clinical settings, according to the post.
Knit joins a run of early-stage clinical AI financings this fall. HealthLeap, which makes software that flags hospital inpatients for review, announced $38 million in seed and Series A funding in October.
What to watch
The company did not name any health system customers or say when its first deployments would go live. The next signals are which providers sign on for its navigation and scheduling uses, and whether outside researchers can test the benchmark claims.



