Perspectives
Our partners write about what we are observing in healthcare AI, legal technology, and financial services automation — evidence-based, operator-informed, and focused on what compliance actually means for building durable software companies.
The FDA's updated framework for AI-enabled devices issued this spring is not a burden on the companies we back — it is a forcing function that separates the serious from the provisional. Here is what it means for builders.
ReadThe first wave of legal AI automated search. The second wave is attempting to automate judgment. Understanding the difference — and the liability implications — matters enormously for anyone building in this space.
ReadTwo years into the ambient clinical documentation era, the patterns are becoming legible. Physician adoption, EHR integration friction, and the economics of note generation have all surprised us in ways worth writing down.
ReadWe have watched three portfolio companies get approached for acquisition in the past 18 months. Every serious buyer asked the same question first: how defensible is the compliance infrastructure? The answer determined the conversation that followed.
ReadPrior authorization consumes an estimated $54 billion annually in administrative cost across US healthcare. The companies solving it are not replacing clinicians — they are removing friction from a process that currently slows down patient care while generating no clinical value.
ReadTwo years after the generative AI inflection point, the fundamental constraint in enterprise legal technology is unchanged: trust, accountability, and defensible outputs. The technology has improved dramatically. The procurement process has not.
ReadRevenue cycle management is not a product category that generates press coverage. It generates margin. For hospital systems operating on 2–4% operating margins, AI that reduces denial rates by 15 points is not a nice-to-have — it is a survival tool.
ReadEmbedded lending and payments infrastructure decisions made at Series A almost always constrain what a company can do at Series C. The compliance architecture is not plumbing — it is the foundation. Here is what we see founders getting wrong, and why.
ReadDiagnostic AI in imaging has attracted extraordinary investment and extraordinary expectations. The gap between what models can do in research environments and what they can do reliably in clinical practice remains consequential. We explore what it takes to close it.
ReadFINRA's 2024 examination priorities letter identifies surveillance deficiencies and recordkeeping gaps as top-tier concerns for the third consecutive year. For compliance technology companies, this is not a threat — it is a mandate letter.
ReadEvery conversation we have with health system CIOs includes some version of the same concern: the models may be good, but can we rely on them when something goes wrong? Trust, auditability, and clear accountability chains are the actual product in clinical AI.
ReadTwelve months into Fund II deployment, we have backed five companies, passed on more than forty, and learned something material about what Series A traction looks like in regulated-industry AI versus what it looked like when we started the fund in 2022. We share that learning here.
ReadLarge language models can now review a standard commercial contract faster than a junior associate and flag the same risk categories with comparable recall. What they cannot do is be held professionally accountable for what they miss. That asymmetry is the entire market.
ReadPCI-DSS version 4.0 came into effect in March 2022, giving merchants and processors until March 2025 to achieve full compliance. The compliance window is a product window for tokenization infrastructure companies whose architecture maps cleanly to the new standard's scope-reduction requirements.
ReadWhen we raised Fund II in 2022, our thesis had evolved in three important ways from when we raised Fund I in 2019. The regulatory environment had shifted, the quality of AI tooling had changed, and we had four years of portfolio learning to draw on. We explain what changed and why.
ReadThe ONC's final rule on information blocking and FHIR-based interoperability APIs created a new floor for health data portability in 2021. Eighteen months later, the effects are becoming visible in how health IT companies architect their platforms and how payers and providers think about vendor relationships.
ReadVertical SaaS companies in healthcare and legal have historically competed on workflow fit and integrations. AI is changing the competitive dynamics in a way that makes the compliance architecture itself the primary differentiator. The companies that understand this first will be very hard to catch.
ReadTwo years of clinical technology deployment under emergency conditions produced three structural changes to how health systems evaluate and adopt software. These changes are permanent features of the landscape, and they favor companies that have already crossed the institutional validation threshold.
ReadEighteen months after leading Casetext's Series A, we have learned things about AI adoption in law firms that we did not know when we wrote the check. Some of what we expected was right. Some of what the market expected was wrong. The delta is worth examining.
ReadKnow-your-customer and anti-money-laundering processes consume an estimated $274 billion in annual compliance cost globally, according to LexisNexis. Automation is overdue. The market is large. The sales cycle into a bank's compliance function is unlike anything in consumer or enterprise SaaS.
ReadHealthcare software founders often treat the business associate agreement as a paperwork milestone. We treat it as a competitive signal. A company that has negotiated and executed BAAs with three hospital systems has crossed a qualification threshold that most competitors cannot easily replicate.
ReadData enrichment started as a sales intelligence product. The companies using it most intensively by 2020 were not sales teams — they were compliance and onboarding functions at fintechs and financial institutions who needed rapid company and counterparty verification without dedicated data sourcing operations.
ReadThe HHS Office of the National Coordinator's proposed rule on interoperability and information blocking, finalized in March 2020, is the most significant structural change to health data infrastructure in a decade. Its effects on investment theses will take years to fully materialize, but the direction is clear.
ReadWhen we closed Fund I in March 2019, the thesis — that healthcare, legal, and financial services would reward AI companies with deep domain expertise and compliance-grade architecture — was contrarian. Eight months later, we are more convinced. Here is why.
ReadIn healthcare, legal, and financial services, a generation of software companies is emerging that treats compliance not as a feature layer but as a core architectural principle. These companies are not selling into regulated industries despite compliance requirements — they are winning because of them. We call this the compliance-native model.
ReadNorthbarn Capital closed its first fund in March 2019. This piece explains what we believe, why we believe it, and what we are looking for in the companies we will back. We write it for founders, for LPs, and for our future selves — so we can hold ourselves accountable to the thesis we started with.
Read