Knit Health has raised US$35m in Series A funding to support the commercial deployment of its Med-BEAM AI model for predicting patient care needs.
The investment brings the healthcare artificial intelligence company’s total funding to US$46.6m and will support the introduction of its technology into clinical and healthcare workflows.
Med-BEAM is designed to analyse medical records and clinical decisions to predict what care patients may require and evaluate how different interventions could affect outcomes.
The company describes its technology as a healthcare “world model”, an AI system trained to understand patient care over time by analysing clinical decisions and medical records.
Unlike general-purpose large language models, which primarily process text, Med-BEAM has been trained on billions of healthcare decisions and hundreds of millions of patient records.
The system examines clinical actions, including diagnoses, treatment orders and referrals, to predict care requirements and the effects of different clinical interventions.
Jonathan Kolstad, co-founder and chief executive of Knit Health, said: “Healthcare still operates too often on what amounts to chance: which doctor you see, which day you arrive, which hospital you enter.
“We’re building the intelligence layer that makes that judgement available for every patient, everywhere.”
Knit Health, founded in 2025 by scientists from the University of California, Berkeley, plans to deploy the technology across hospitals, health systems, life sciences companies, healthcare technology providers and AI developers.
Potential applications include care management, patient navigation, clinical trial recruitment, specialist scheduling, hospital-at-home services and predicting responses to treatment.
The company reports that internal benchmarking showed Med-BEAM outperforming several frontier AI models when predicting clinical actions for hospitalised patients over a 24-hour period.
According to Knit Health, its model was up to four times more accurate, 100 times faster and 250 times less expensive than the models included in the comparison.
The company also reported better prediction calibration, meaning the model’s confidence levels were more closely aligned with observed outcomes.
However, the benchmark results were reported by Knit Health, and the announcement did not identify the competing AI models or provide details of independent validation.
The company says its training data is fully de-identified before being received, meaning information that could identify individual patients has been removed.
It also states that it uses only datasets for which it has documented rights.
Med-BEAM incorporates confidence calibration designed to identify predictions that lack sufficient supporting evidence.
The Series A funding round was led by New Enterprise Associates (NEA), with participation from Town Hall Ventures, SignalFire and existing investors.
Blake Wu, partner at NEA, said: “One underlying model can power dozens of critical workflows from care management and patient navigation to clinical trial recruitment, specialist scheduling, hospital-at-home, and treatment-response prediction.”