One of healthcare’s most promising early use cases, AI scribes, is under pressure as Epic recently announced their homegrown product after years of partnership with Abridge.

The update comes at a time when the sector faces crushing administrative burdens, with 20–25% of total annual U.S. healthcare spend going toward administrative tasks. Meanwhile, 49% of doctors report symptoms of physician burnout.

This backdrop presents a clear, urgent opportunity to automate and streamline a wide range of processes. Generative AI has the potential to boost revenue, reduce costs, expand access to care, and improve the patient experience. While exciting AI healthcare applications have emerged over the last two years, EMRs are racing to offer similar capabilities directly to their providers. At Bain Capital, we’ve had a front row seat to the EMR evolution in healthcare through our work with Athenahealth.

Navigating this ecosystem is complex. Most successful companies will have some EMR reliance, but the question is how to thread that needle while building an enduring business.

Build alternative workflows or data access and focus on the long tail

The EMR is a treasure trove of data, and it’s easy to rely on that data alone to build an application that can deliver tremendous value for providers. Let’s use pre-visit chart summaries as an example: many providers we’ve spoken with say it would be valuable to receive a brief summary of a patient’s chart before a visit. A simple solution might take a patient’s chart from the EMR and summarize it using AI. A more defensible solution might add additional workflows to acquire patient data from other sources such as specialist visits, labs, or the fast-growing plethora of consumer health tools like Function or Levels.

There are also workflows that bridge data across platforms, EMRs, payers and patients. For example, prior auth automation platforms rely on external, non-EMR data. Portfolio company Silna Health leverages data from payers to complete insurance eligibility and verification, and navigate payer-side workflows to submit prior authorizations. Strengthening your capabilities and integrating with external data sources makes it harder for EMRs to copy your product, while increasing buyer loyalty by becoming their trusted source for accurate, complete access to a critical data set.

Most AI healthcare platforms quickly chase after EMR partnerships to deeply embed into a provider’s workflow. This has a ton of benefits: ease of adoption in a provider’s native environment, distribution acceleration through partnership, and streamlined integration. In outpatient, this strategy can be a strength, not a weakness, given the market landscape of EMRs is more fragmented.

The outpatient market tends to have more of a marketplace approach, where they have partner vendors that offer specialized tooling to their customer base. They cover a wider variety of specialties, size or organization, use case, etc. relative to systems.

Create an unmatched offering

Abridge is deeply embedded within Epic, but Epic’s newly announced homegrown tool could change the dynamic. Abridge would be harder to replace if its platform were viewed as independent from the EMR—both in how users learn and engage with it, and in the unique product capabilities it offers that Epic can’t match.

A pattern is emerging that we’ve seen in many other industries. Just as tech behemoths have, for decades, observed newer tech and replicated its features, so too will the biggest healthcare platforms. They’re advantaged because they already have distribution.

Competing as a software tool selling to providers means creating a unique product capability, data advantage, or interface that’s hard to replicate, or focusing on segments where these vendors have less control.