Knit Health Launches with $11.6M to Build AI From Clinical Decision Patterns
Knit Health, a UC Berkeley spin-out, launched from stealth with $11.6 million in seed funding to deploy artificial intelligence trained on how clinicians actually make decisions. The round was co-led by Uncork Capital and Frist Cressey Ventures, with participation from Moxxie Ventures and Coalition Operators.
The company built its Large Clinical Behavior Model using data from 130 million patient records across 30 U.S. health systems. Rather than training on published medical literature like most healthcare AI systems, Knit's model learns from observed patterns in referral decisions, scheduling choices, and how clinicians navigate institutional constraints.
What Sets Knit Apart
Most AI for healthcare relies on language models trained on textbooks and research papers. This misses what actually drives better outcomes: the informal knowledge embedded in how experienced clinicians route patients, coordinate care, and make decisions under real-world constraints.
Knit's approach differs in three ways:
- Learns from behavior, not text: Using deep reinforcement learning and causal inference, the model captures decision sequences as they unfold in practice rather than generating probabilistic text responses.
- Customized to each health system: Knit fine-tunes to a specific organization's workflows, capacity limits, and referral patterns, allowing it to integrate into existing operations without requiring staff retraining.
- Operates as infrastructure: The system sits beneath routing decisions, discharge predictions, care team assignments, and referrals-touching every workflow that affects a patient.
Initial Deployments and Governance
Knit Health is deploying initial models for triage, patient flow, and quality improvement across partner health systems. The company built in HIPAA compliance, bias testing, and continuous monitoring from the start.
Jonathan Kolstad, co-founder and CEO, said the model captures knowledge clinicians develop across millions of patient journeys. "Much of what matters most in medicine isn't written in textbooks," he said. "It's learned through experience navigating the healthcare system."
Understanding generative AI and LLM applications helps healthcare professionals evaluate how these systems can fit into existing workflows and what governance structures are needed.
Your membership also unlocks:
