Last Updated: 12 May 2026
Knit Health, a startup spun out of the University of California, Berkeley, has raised $11.6 million in seed funding and stepped out of stealth. The company is building AI trained on real clinical behavior rather than textbooks or published research. Its founders, a team of Berkeley researchers and academics led by CEO Jonathan Kolstad, call the technology a Large Clinical Behavior Model, or LCBM. In plain terms, it is a system that learns from the patterns doctors follow when making decisions, not from written medical knowledge.
Most healthcare AI today can read. It cannot observe. Tools built on large language models are trained on text. They can pull information from a patient chart or suggest a likely diagnosis, but they have no sense of how a hospital actually runs. Why does one patient get routed to a specialist while another does not? How does a discharge decision in the morning affect bed availability by afternoon? These are the kinds of questions that existing tools cannot answer, because the knowledge that drives them was never written down anywhere. Healthcare AI spending reached $1.4 billion in 2025, nearly three times what was invested the year before, yet much of that money is still funding text-based tools that replicate what is in books rather than what happens in practice.
The seed round was co-led by Uncork Capital and Frist Cressey Ventures. Moxxie Ventures led the earlier pre-seed round, with Coalition Operators also participating.
The problem Knit is working on is specific. It is not about making AI smarter in a general sense. It is about capturing the kind of practical knowledge that experienced clinicians build over years of seeing patients, knowledge that exists only in behavior, not in any document or database.
The AI-powered clinical decision support market is projected to grow from $857.5 million in 2025 to over $2.3 billion by 2032. Health systems are fueling that growth as they look for tools that reduce mistakes, speed up workflows, and help doctors make better calls faster. Most products chasing that opportunity, though, are still built on the same foundation: static data sets and probabilistic text outputs. Knit is taking a different technical approach.
"The hardest challenge in healthcare isn't knowing what good care looks like, it's delivering it consistently for every patient. Knit Health's model embeds the best clinical intelligence directly into the workflow, helping clinicians make better decisions faster and more consistently. At scale, this will improve patient outcomes and transform clinical operations across health systems."
Navid Farzad, Managing Partner at Frist Cressey Ventures
The new capital will go toward further developing the LCBM and expanding partnerships with health systems. Knit is currently deploying its first models in three areas: triage, patient flow, and quality improvement. These happen to be some of the most complex operational problems any hospital faces, and the ones where getting decisions wrong carries real costs.
Longer term, the company wants its system to sit underneath every routing decision, discharge prediction, care team assignment, and referral that a health system processes. That is a large ambition for a company in its first year. The architecture, the company says, is being built with that kind of scale as the target.
Knit Health was founded in 2025 by researchers from the University of California, Berkeley. The founding team brings together expertise in behavioral economics, causal inference, generative AI, and healthcare operations. Jonathan Kolstad serves as CEO. Maya Petersen, MD, PhD, is Co-Founder and Chief Scientist. Jonas Knecht and Anshul Amar serve as Co-CTO, with Knecht also holding a Co-Founder title. Ted Robertson is Co-Founder and COO. Sophie Pinkard leads product and Midori Uehara leads growth.
The LCBM is trained on electronic health record data from over 130 million patients across 30 U.S. health systems, sourced through Truveta. Rather than learning from text, the model uses deep reinforcement learning, causal inference, and behavioral cloning to study sequences of clinical decisions. It then fine-tunes to each health system's own patterns, which means it adjusts to how that specific institution operates rather than applying a one-size-fits-all layer on top of existing workflows.
"Much of what matters most in medicine isn't written in textbooks, it's learned through experience with time and navigating the healthcare system. Across millions of patient journeys, clinicians develop patterns for what to do next and when. Knit learns from those real-world decisions, transforming collective clinical experience into intelligence that improves how the system works."
Jonathan Kolstad, Co-Founder and CEO of Knit Health
The system is built to comply with HIPAA standards and includes governance protocols, bias testing, and continuous monitoring.
Uncork Capital and Frist Cressey Ventures co-led the seed round. Moxxie Ventures led the pre-seed, with Coalition Operators participating in that earlier raise. Knit Health has not disclosed any other institutional funding outside these two rounds.
"Knit Health is creating a new approach to AI. Unlike traditional models, it learns and evolves from real human behavior and can be applied across complex systems. This approach redefines how intelligence is captured and scaled, opening entirely new possibilities for AI-driven innovation in healthcare."
Tripp Jones, General Partner at Uncork Capital
