For years, the NHS debate around Epic has been about the electronic patient record: implementation cost, clinician adoption, interoperability and whether a large US platform should sit at the centre of a British hospital. That debate is rapidly becoming outdated. Epic’s latest AI strategy suggests it no longer wants to simply store the NHS clinical record. It wants to become the intelligence layer through which clinicians work, patients interact, hospitals operate and increasingly, decisions are made.

That matters because Epic’s footprint inside the NHS has been expanding at exactly the moment NHS England is pushing hospitals towards near-universal electronic patient records. NHS England says £1.9 billion has been invested through frontline digitisation, with 95% of trusts expected to have implemented or upgraded an EPR by March 2026. By March 2025, 91% had already implemented one. Epic therefore arrives at its next phase with a potentially powerful advantage: it is already embedded inside the clinical workflow of some of the NHS’s most digitally influential organisations. The next battle is not necessarily about winning another EPR procurement. It is about how much more of the hospital can be absorbed into the platform once Epic is already there.

The scale of that ambition became much clearer this year. Epic is building AI around three main interfaces: Art for clinicians, Emmie for patients and Penny for administrative and revenue-cycle work. Behind them sits Agent Factory, which allows organisations to create AI agents capable of reasoning and acting across workflows, while Curiosity, Epic’s family of healthcare foundation models, is trained using its Cosmos dataset of more than 300 million deidentified patient records. Epic says more than 85% of its customers are already using some form of Epic AI. This changes the proposition dramatically. An EPR traditionally waits for somebody to enter, retrieve or interpret information. An agentic platform can potentially identify work, interpret context, initiate actions and coordinate the next step. Epic is already describing agents being used for areas such as infusion preparation, discharge processes and clinical workflows. In an NHS hospital this could eventually touch referrals, outpatient preparation, waiting-list validation, coding, discharge, medicines, scheduling and patient communications.

This arrives at an almost perfect moment for Epic. NHS England is explicitly asking providers to deploy ambient voice technology, increase use of AI-assisted services, achieve complete EPR coverage and make greater use of digital platforms to improve waiting-list management and productivity. If Epic can offer these capabilities inside software clinicians already use, the attraction for trusts is obvious. There is less integration, fewer separate logins and potentially a much faster route from AI experiment to operational deployment.

But this is also where NHS leaders should become much more sophisticated buyers. Every new workflow placed inside the EPR increases the cost and complexity of ever leaving it. The question should therefore shift from “Should we buy Epic?” to “Which parts of our future operating model should Epic be allowed to own?” Hospitals need strong interoperability, open APIs, transparent AI governance and the ability for specialist technology companies to operate alongside the core record. Otherwise the NHS could spend years arguing about EPR vendor lock-in only to recreate it at a much deeper level through AI.

There is another strategic issue. The NHS is simultaneously developing a Single Patient Record designed to bring information together across organisations. That makes architectural choices increasingly important. The future NHS cannot simply become a collection of extraordinarily sophisticated digital islands, even if each island works brilliantly.

Epic’s move into AI is therefore potentially very good news for NHS productivity. It could remove enormous amounts of administrative work and make clinical data dramatically more useful. But it also changes the balance of power. Epic is no longer competing merely to be the software that records healthcare. It is competing to become the platform that helps decide what happens next.

For the NHS, that is a much bigger decision than buying an EPR.