Northbarn Fund II: Where We Stand After Twelve Months of Deployment
Twelve months into Fund II deployment, we have backed five companies and passed on more than forty. That ratio — five funded, forty-plus passed — is approximately what we expected. Regulated-industry AI at Series A is a market where the volume of interesting companies is high and the number that have crossed the institutional validation threshold we require is lower. The funnel is appropriate for the thesis.
What surprised us in the first year: the quality of companies coming into the market with genuine prior authorization and revenue cycle automation products was higher than we expected. The pandemic's impact on health system financials — negative operating margins for many systems in 2020 and 2021, followed by recovery pressure in 2022 — created acute motivation for health systems to evaluate AI-driven administrative automation. Companies that could demonstrate 15-to-20-point improvements in prior authorization approval rates or denial reversal rates were finding health system CIO attention more readily than we had modeled.
What did not surprise us: law firm adoption of generative AI contract review tools was faster than the conventional wisdom suggested but constrained by the professional responsibility questions we had already identified. The firms adopting fastest were doing so with explicit supervision protocols — attorney review checklists, billing code guidelines for AI-assisted work, and malpractice insurance rider updates. The firms adopting slowest were waiting for state bar guidance that in many jurisdictions has still not been issued.
Looking ahead to the remainder of Fund II deployment: we are tracking generative AI applications in financial services compliance monitoring closely. The maturation of LLM capabilities for document analysis — processing regulatory filings, examination letters, and policy documents — is creating a genuinely new capability for compliance functions at banks and asset managers that did not exist two years ago. The companies building here face the same trust and auditability requirements as legal AI, with the additional complexity of examination-ready output requirements. We believe this is a deep and durable market and are actively evaluating investments in it.