Rethinking EHRs in an AI-first World

Sam Toole is an entrepreneur and operator turned investor with NYC-based Primary Ventures.

Since 2020, >$30Bn has flowed into verticalized electronic health records, largely through private equity transactions and acquisitions. These deals rarely show up on venture capital radars, but they signal two important things. 

First, across most specialties, clinicians are still running core workflows on systems built more than a decade ago. Second, the largest software outcomes in healthcare over the past decade are generally not venture-backed! And as a result, we’re missing them. 

Select Recent EHR Transactions (non-exhaustive)

As a healthcare investor for the past 10 years, I’ve been thinking more broadly about what the future of application software for medicine looks like and how AI will reshape software-as-a-service or SaaS. Software-as-a-service refers to cloud-based models where companies host applications and then deliver them over the Internet on a subscription basis.

There’s already plenty of debate out there amongst my fellow investors about systems of record becoming systems of action. A system of action refers to a new generation of enterprise software, primarily powered by AI, that moves beyond storing and managing data to actively performing work and automating complex workflows. This piece is not to debate that future. Instead, I want to imagine a future where existing systems of record are paired with, or in some cases replaced by, true systems of action. In healthcare, EHRs, which more than 90 percent of doctors use every day, are the most natural place to start. I will be posting a handful of pieces on this theme in 2026 that go deeper into the market, the product, and potential business models shift enabled by AI. 

While most investors view investing in the EHR market as an opportunity from twenty years ago, I view it as a market that is ready to be disrupted. With advances in AI and physicians’ increasing willingness to adopt new technology, there will be new category-defining winners here.

Do I believe that there will be a crop of new start-up EHRs in every specialty? Probably not. The process of ripping out and replacing any system of record is brutal. Most providers genuinely dislike their EHRs, but they live inside them every day. The change management alone is enormous, and the perceived risk often outweighs the promise of incremental improvement. The standard venture playbook has been to build a point solution, or wedge, on top of an existing system of record, earn trust, and then slowly eat away at the core platform before (hopefully) making the swap. I’m not convinced. As a venture capitalist, when an early-stage founder pitches me the idea of a system of record replacement, it’s generally met with a huge amount of skepticism. Watching how the “AI scribe wars” are unfolding makes me skeptical that the traditional wedge approach will work today. Incumbents are willing to copy anything that starts to matter. See Epic, Athena, and many more examples.  

Epic is the most egregious example of how quickly momentum can be shifted.  For years, the Abridge-Epic partnership was the engine behind Abridge’s breakout growth. Abridge is one of the most well-funded and scaled AI scribes, now used by tens of thousands of doctors across the world. Epic’s Workshop program effectively created a sanctioned, first-of-its-kind path for Abridge to build deep integrations, co-sell alongside Epic, and access early customers inside the Epic network. By all accounts, that early integration and co-marketing campaign was critical.