Most of us have had the same experience in a car. You try the inbuilt sat-nav: expensive, impressive and fully integrated into the dashboard. But within minutes you realise that the maps are outdated, the traffic data is wrong and the routing is clunky – so you ignore it and open Google Maps on your phone.
The manufacturer spent millions building its own navigation system, but drivers chose the app that actually works. The NHS is at risk of making the same mistake with intelligent navigation.
Faced with this new technology, the NHS’s instinctive response could be to tinker with the triage and navigation system it already has: making marginal upgrades to 111 and the NHS Pathways algorithm that sits behind it. It would be a large procurement, take years to implement and inevitably cost a great deal of money, but if it didn’t fundamentally improve the consumer experience, it would be profoundly useless. Patients would simply route around it, using AI assistants on their phones instead.
This is not hypothetical. Right now, millions of people are turning to the likes of ChatGPT – general-purpose AI assistants – for health advice. They’re uploading GP letters, symptom histories and test results. They’re asking, “Should I see a doctor?” “Is this urgent?” “What does this blood test mean?” And they’re getting answers that are fast, personalised and – crucially – good enough (or, at least, that are perceived to be).
These aren’t medical AI systems. They haven’t been trained on clinical data and they’re not regulated as medical devices. But they’re being used for health care anyway, because they solve a problem that the NHS has failed to address: how to navigate an impenetrable and impossibly complex system.
This matters more than most health leaders realise. Because what looks like a convenient consumer tool today could become the greatest threat to the founding principles of the NHS, should the system fail to engage.
The “Doctor in Your Pocket” Is Already Here
OpenAI and Anthropic didn’t set out to build health-navigation tools: they built general AI assistants that happen to be remarkably good at understanding health information. These systems operate within secure user accounts, can process personal medical records and provide guidance that – while imperfect – is far more accessible than trying to get through to your GP surgery at 8am.
People are pragmatic. They tolerate imperfection in exchange for speed and convenience. And these assistants are extraordinarily convenient: available 24/7, infinitely patient, capable of explaining medical jargon in plain English and improving with every interaction.
The result? These platforms are becoming the default first point of contact for health concerns. Not 111, not the NHS website and not your GP reception, but an AI assistant that knows your entire health history and can respond in seconds. And because capacity is determined by bandwidth, not headcount, these services are infinitely scalable too.
Data Gravity Creates Platform Power
Here’s where it gets dangerous for the NHS. Because these assistants are so useful, people keep feeding them more information. Every GP printout, every outpatient letter and every symptom update adds to a growing, longitudinal health record that exists entirely outside NHS systems.
These platforms are building something that the NHS has always struggled to create: a complete, structured, enduring history of an individual’s health. They don’t just know what’s been coded in your GP record, but everything else too: the back pain that comes and goes, the anxiety you’ve never mentioned to a doctor, the family history you’ve shared (and haven’t had to repeat ad nauseam).
This creates genuine clinical value in the form of better risk stratification, personalised triage and dynamic symptom tracking. The assistant effectively becomes a private digital health record – one that’s often more complete and accessible than anything currently within NHS systems.
Before too long, you won’t want to use another navigation tool. You certainly won’t want to start from scratch explaining your history to 111. The switching costs – in terms of both data and memory – are too high.
This is classic platform capture. And it’s happening right now.
Triage Shapes the System
Most people think of triage and navigation as a service. And that’s true, but it’s also the NHS’s resource-allocation engine.
Triage helps to determine who goes to the pharmacy, who sees a GP and who goes to A&E; it establishes who waits and who gets prioritised. It also shapes demand, influences utilisation and controls how pressure flows through the system.
If a private platform controls triage and navigation – even if it’s not formally partnered with the health service – it effectively controls those decisions. And platforms don’t optimise for system sustainability or even safety: they optimise for consumer satisfaction and engagement.
Without NHS involvement, these platforms will set their own risk thresholds. They’ll define what “urgent” means according to their liability models and user expectations, not clinical guidelines or system capacity. They’ll route people to wherever is most convenient, which could mean A&E, private digital GP services or subscription health care.
If citizens stop using 111 and ask ChatGPT where to go instead, the NHS loses the ability to shape demand digitally. National guardrails will get bypassed. The system’s capacity to “shift left” will be undermined. And the fundamental ability to manage a universal service that is free at point of use will be compromised.
The Insurer Stack Is Being Built
Globally, health insurers understand this dynamic – often better than providers do. That’s why they’re racing to own triage and navigation: it’s the control point that determines downstream costs and utilisation patterns.
Look at what these AI platforms already have: the consumer interface, the data layer and increasingly sophisticated risk stratification. All they need now are some contracts with providers and a monthly subscription and, before you know it, they’re a health system.
At that point these platforms would effectively have the insurer stack. The NHS would become a back-end supplier of services, disintermediated from the citizen relationship that it was built to serve.
Digital primary care could easily become a loss leader – a way to acquire and lock in users before routing them through a vertically integrated private health-care system. If platform capture were to be completed, the NHS would simply become one option among many, competing on convenience rather than principle.
A Fork in the Road
The NHS has a choice: it can ignore this shift and watch platform capture accelerate, or it can engage strategically and shape how these tools work safely, effectively and within a universal health-care context.
If the NHS engages, it could lean into the technology and add to the consumer experience. It could feed in real-time information about which A&E departments are full or which have the shortest wait. It could schedule a GP appointment if you need one or prescribe you something if self-care would be sufficient. With access to a hyperlocal Directory of Services it could direct you to a specialist neighbourhood provider that is suited to your needs, rather than defaulting to traditional GP surgeries and hospitals.
Similarly, the NHS could work with LLMs to bolster the safety of these services, embedding NHS-specific guardrails and risk thresholds. This would maintain public control over the mechanism that determines who gets care – and when and where they get it. The NHS could work with suppliers of both Bayesian decision-support tools and medical LLMs to integrate generative and deterministic AI, thereby preserving safety as well as convenience.
More ambitiously, the NHS could incentivise citizens to contribute unstructured health data to a public infrastructure. This would help to build a population-level data asset that could be used for research and service improvement, with potential for monetisation through partnerships to subsidise care costs, as the Health Data Research Service proposes.
But, critically, this isn’t about productivity gains or doing more with less. It’s about preserving the founding principles of the NHS: universalism, risk pooling, care according to need rather than ability to pay, and free access at the point of use.
If triage and navigation – the mechanisms that make those principles operational – end up being controlled by private platforms optimising for profit, those principles become impossible to sustain.
A Changing Mindset
The current dogma is that AI is a tool the NHS can selectively deploy – on its own terms, in its own time and within existing institutional boundaries.
But AI isn’t just a tool: it changes the rules of engagement for the NHS. AI is collapsing the information asymmetry that has traditionally underpinned modern medicine, meaning patients can interpret test results, understand risk and challenge decisions with a level of fluency that would have required clinical training a decade ago. AI does not simply make services more efficient: it puts patients in charge.
Systems that recognise this – and design around empowered, information-rich citizens – will remain relevant. Systems that cling to paternalistic, controlling models of access will find themselves bypassed. You cannot force people to use the inbuilt sat-nav if Google Maps is better.
If the NHS produces a marginally improved version of 111 while global AI platforms continue to offer seamless, personalised navigation in your pocket, citizens will choose the latter. Not for ideological reasons, but for practical ones.
That is why partnership (not protectionism) is the strategic response. The NHS does not need to outbuild big tech, but rather shape how these tools operate within a universal health-care framework. That means sharing NHS data (even in a sandbox environment) to help sharpen its diagnostic models, embedding NHS safety guardrails, integrating real-time capacity and ensuring that routing decisions reflect public values rather than purely commercial incentives.
Intelligent navigation isn’t a nice-to-have digital upgrade: it’s the battleground on which the future of the NHS will be decided. Either the NHS disrupts itself and maintains control of the front door, or big tech will do the disrupting and the NHS will find itself locked out of its relationship with the people it was created to serve.
The alternative is not stasis but disintermediation. The platform capture of health care has already begun, so the question now is will the NHS fight to remain relevant? Or will it stay wedded to outdated tech and accept its relegation to back-office provider in someone else’s health-care system?