The Digital Strategy of St George's, Epsom, and St Helier: When the Ambition Outpaces the Wiring Acknowledges the NHS

Buried in the board papers of St George's, Epsom and St Helier University Hospitals and Health Group is a sentence that says more about the state of NHS technology than any of the artificial intelligence pledges that surround it. Between £23 million and £25 million of capital in 2026/27 will go on risk remediation, including network hardware and wifi coverage. Not ambient documentation. Not predictive analytics. Wifi.

The group's digital and data strategy to 2030 is, on its own terms, an impressive document. It promises a single electronic patient record properly optimised across sites, a unified data platform, an AI Centre of Excellence, and a target of ninety per cent of patient interactions running through the NHS App within three years. Read the executive summary and you would think gesh is closer to Silicon Valley than to south west London. Read the section on where the group actually stands and a different picture forms: over two hundred systems with limited interoperability, ageing infrastructure, weak networks, and a board assurance framework that has held its digital risk score at twenty despite genuine progress on the shared record. Chief executives across the health service have grown practised at writing both paragraphs into the same document without appearing to notice the tension between them.

This is not a criticism unique to one trust group. Look across recent strategy publications and the pattern repeats with striking consistency: six broadly identical strategic pillars, a foundations-first roadmap, a Copilot pilot, an ambient voice rollout, and a digital front door built around the NHS App. Lewisham and Greenwich has set out 2027 as the year its EPR finally goes live alongside ambient voice technology. Southampton has automated outpatient administration while continuing its own record progress. Each trust writes its version of the same story because each is negotiating the same structural problem, a health service asked to build an intelligent, data-driven future on infrastructure that in many cases predates smartphones.

The political backdrop makes the mismatch sharper rather than softer. Chancellor John Healey is understood to be weighing greater borrowing flexibility for NHS capital ahead of his Budget in late October, against an estate where the bill for high and critical maintenance has very nearly tripled since 2015 to reach almost £16 billion. Delegated spending limits have been raised so that individual capital projects can proceed without Treasury sign-off below £300 million, a genuine loosening on paper. Whether it survives contact with a fiscal event competing against defence and energy commitments is a separate question, and one that determines whether risk remediation budgets like gesh's are adequate or merely a down payment on a much larger repair bill.

None of this means the AI ambitions in these strategies are empty. Ambient voice technology genuinely is reducing documentation burden where it has been deployed properly, and NHS trusts do have credible examples of AI cutting waiting times and easing administrative load. But there is a governance cost to building fast on foundations that are still being poured, and it is already visible. Complaints teams at several trusts are reportedly contending with a rise in lengthy, AI-generated correspondence containing fabricated legal claims, a small but instructive sign of what happens when a technology diffuses through a system faster than the structures meant to manage it.

For NHS leaders the practical lesson is one of sequencing and honesty rather than ambition. A digital strategy that leads with an AI Centre of Excellence but spends its first year's capital on cabling is not being duplicitous, it is being realistic about where the actual constraint sits, and boards would do well to say so more plainly rather than let the AI section carry the document. For policymakers and the Treasury, the gesh figures offer a rough exchange rate between digital rhetoric and digital reality, roughly twenty five million pounds just to make one group's network reliable enough to run the systems it already has. Multiply that across the trust landscape and the scale of what "data-driven NHS" actually requires becomes considerably less abstract than it sounds in a foreword.

The strategy documents will keep arriving through the autumn, each promising broadly the same connected, intelligent future. The trusts worth watching are the ones prepared to admit, as gesh has, that the wifi has to work first.