Knit Health Launches With $11.6M to Build Clinical AI From Real Patient Data
Knit Health, a health tech startup spun out of UC Berkeley, launched today with $11.6 million in seed funding to deploy artificial intelligence that learns from how clinicians actually make decisions. The round was co-led by Uncork Capital and Frist Cressey Ventures.
The company has built what it calls a Large Clinical Behavior Model (LCBM)-trained on electronic health records from over 130 million patients across 30 U.S. health systems. Rather than relying on published medical literature or text-based training, the model learns patterns from real clinical workflows: how doctors refer patients, schedule appointments, and navigate institutional constraints.
Why This Matters for Healthcare Operations
Most AI for Healthcare today trains on textbooks and research papers. That approach misses what actually drives better outcomes: the collective knowledge clinicians build through experience about which patients need what care, when.
Knit's model learns directly from clinician decision-making patterns using deep reinforcement learning and causal inference. It then applies those insights to optimize patient routing, discharge planning, care team allocation, and referrals within specific health systems.
The company fine-tunes its model to each health system's practice patterns and capacity constraints, rather than deploying a one-size-fits-all solution. This means the system integrates into existing workflows from day one.
How It Works
Knit's approach differs from traditional clinical AI systems in three ways:
- Behavioral modeling: The model learns from sequences of actual clinical decisions, reflecting how care unfolds in practice rather than generating probabilistic text.
- Health system-specific: It adapts to each organization's unique referral dynamics and operational constraints.
- Infrastructure layer: Knit sits beneath every routing decision, discharge prediction, and care team assignment-making it foundational to clinical workflows rather than a standalone tool.
The company is built with HIPAA compliance, bias testing, and continuous monitoring built in.
What's Next
Knit Health is partnering with health systems to deploy initial models for patient triage, flow optimization, and quality improvement. The founders-researchers from UC Berkeley with expertise in behavioral economics, causal inference, and generative AI and LLM technology-say the goal is to make Knit the foundational infrastructure layer that clinical decisions run on.
Your membership also unlocks:
