Three forces are converging at once. The ten-year health plan is driving three major shifts, from hospital to community, analogue to digital, sickness to prevention. Structural reform is redrawing institutional boundaries and consolidating governance. Elsewhere, technology funding is rising by around 50%, to £10bn by 2028/29. It is cheaper and more effective to build AI into systems as they are being designed, rather than to bolt it onto established ones. Change is easier when the system is already in motion.
Done well, AI eases the operational and financial pressures that continue to stretch the system. It also addresses the three fronts that matter most to Boards: better patient outcomes, higher workforce productivity and lower cost of care. This triple dividend could lead to a 1% annual cost reduction, a 2% productivity improvement and ensure the 92% 18-week treatment standard. But the NHS is a different organisation to most, so AI strategies designed for the private sector do not translate directly.
Redesign around AI for success
Bolting AI onto existing pathways as a point solution does not work. It simply automates inefficient legacy processes, adds complexity, traps value in silos and creates governance sprawl. It can only deliver its full value when it replaces a legacy step, and the workflow is redesigned around it. This must be the starting point for every AI investment a Trust or ICB makes.
Take Mid and South Essex NHS Foundation Trust, for example. By redesigning the appointment workflow around Deep Medical's AI rather than overlaying it, the Trust cut DNAs by 30%, enabled 152,000 additional patients to be seen, and released £27m of funded capacity annually.
“Implementing AI onto an existing pathway never realises the benefits. You must redesign the workflow with AI in mind, in order to get the most value.”
Principal Clinical Scientist, NHS Trust
Fix the foundations to drive real progress
Five dimensions need to move together: foundational infrastructure and capabilities, strategic enablement, cultural transformation, operational redesign, and governance and oversight. Of these, the foundations matter most. Without them, benefits stay local and never scale.
Progress is already underway: EPR adoption is approaching 100%, the NHS App is used by over a quarter of the population monthly, and the Federated Data Platform has engaged 150 Trusts. Yet just 32% of Trusts have reached digital maturity, 68% of doctors don't believe the NHS has the right digital infrastructure, and only 6% of doctors have access to AI training despite 79% saying they need it.
Pick the right place to start
Seven characteristics will define where AI fits best: well-defined workflows, high-volume repeatable actions, reliable data, self-learning potential, moderate precision with safety guardrails, governed and traceable actions, and clear decision rights. Diagnostics, ophthalmology and dermatology emerge as the strongest clinical candidates, though the back and middle office score equally well on the same criteria, having so far been largely untouched in the NHS.
Investment should sequence across three overlapping phases. Foundations first, then quick wins that deliver measurable ROI within existing budgets, followed by transformational capital for higher-complexity use cases once the evidence base, infrastructure and workforce confidence are in place.
The biggest long-term prize is the opportunity to optimise finite resources across the whole system. The data architectures, governance models and operating models built in the next 18 to 36 months will determine whether that prize is realistic.
“We must think through what the NHS should look like in five years' time, and plot the path toward it, knowing that tech and AI will drive the most fundamental change.”
CFO, NHS Foundation Trust
No time to wait: Why AI transformation in the NHS starts now
Contact us
