The NHS is preparing for a decade-long transformation of its digital foundations, data architecture and care-coordination processes, driven by agentic AI and next-generation interoperability, according to Kartik Taneja, Business Intelligence and Automation Lead at the NHS.
Addressing delegates at the first-ever QA Financial Healthcare & Insurance Forum London 2025, Taneja outlined a roadmap that places AI, testing maturity and federated data models at the centre of a more proactive and connected healthcare system.
Taneja emphasised that the NHS remains one of the most misunderstood digital ecosystems in the world. “It is not a monolithic organisation. You can think of it as a federation of organisations, more than 100 organisations to be accurate.”
These range from GP practices and hospitals to community care, mental health providers and administrative bodies. “So the idea of telling you all this was we are looking at a relatively complex landscape of organizations, among other things.”
AI now features explicitly in the NHS’s long-term digital strategy. “In our last 10-year long-term plan, there’s a specific focus on AI as to how we can elevate various challenges we are facing by using AI within NHS.”
The pressures driving that shift are immediate. “The context is NHS, as most of us might know, is at a sort of a turning point. There is rising demand. There is workforce constraints, especially on the medical sector.”
And AI itself is reaching maturity. “AI is maturing to a situation where they can be put into production use cases,” he said. “And AI also supports our strategic shift that we are undergoing from acute care to community care and towards prevention.”
To operationalise that shift, the NHS has begun creating digitally supported neighbourhood teams. “What we have started to do now to operationalize all these concepts is we are setting up neighborhood teams with support of AI processes, augmenting the care process,” Taneja said.
But today’s systems remain fragmented. “Currently, we have a landscape of very siloed systems, so each of these NHS organizations would have their own IT system, which does not talk to an IT system outside with other NHS organisations.” Too many processes still depend on phones, emails and traditional letters. “It leads to an incomplete view of patient data,” he said.
“We don’t know which treatment you went for maybe one or two years back… that limits our ability to provide timely interventions.”
Neighbourhood teams are designed to bridge those gaps. “The core members of these neighborhood teams are the general practitioners, the GPs, the nurses, the social care worker, mental health practitioners, and pharmacists,” he said.
“They would typically be catering to a patient demography of 30,000 to 50,000 patients.”
Their purpose is simple: “Break the barriers between organizations so that seamless care can be provided… in a timely and proactive manner.”
Such teams require precise, reliable and interoperable data flows, issues that sit at the heart of QA and software-testing challenges across the NHS.
This is where agentic AI enters the picture. “Agentic AI refers to any systems that can perceive the environment, make independent decisions, take actions to achieve specific goals, and learn from the outcomes,” he said. “So all these four elements are what really defines an agent or agentic AI.”
He divided healthcare agents into four classes: those supporting patient monitoring, those supporting clinical decision-making, those managing care plans and handovers, and those that optimise system-wide capacity. Each depends on accurate, well-tested data pipelines, a theme resonant with QA Financial’s audience.
Taneja connected these agents to the NHS’s broader multi-year digital architecture. “From a data architecture perspective, we first get to a unified patient view, then we get to a sort of a data mesh where the care plans come together, and then finally we have the ideal system, intelligent platform, which can do the data and the agency processes together.”
The initial steps are already deployed. “We currently have implemented federated data platforms, where we’re pulling in data from multiple systems. But it’s not perfect, but first step is there.”
“AI supports our strategic shift that we are undergoing from acute care to community care and towards prevention.”
– Kartik Taneja
Taneja argued that AI is essential not only for clinical outcomes but for the entire testing and integration pipeline underpinning NHS data systems.
“AI can basically connect with these multiple systems through APIs and enable those workflows and the data information needs for these teams quite quickly,” he said. “The clinicians are involved in co-designing these tools, and we use the FHIR standard to connect basically.”
Taneja described a forward-looking use case: early detection of diabetes. “The markers are there even three, four years before you are diagnosed as diabetic,” he said.
“If you can turn that around at that point in time, that’s a big win.” A more advanced vision includes digital twins. “Your entire history of tests are basically put in a sort of proactive data architecture,” he said.
“A clinician can ask this question, if you are becoming diabetic, what is the best treatment that can be provided to you to prevent the occurrence of diabetes… based on the historical data that’s available in other systems.”
Even the commissioning model would change. “Currently we have what we call an activity-based commissioning,” he said. “But we would like to transition to outcome-based commissioning. Basically what it means is how many illnesses they prevented.”
Throughout the talk, Taneja emphasised the need for careful testing, validation and explainability. “The AI has to be explainable in terms of why a particular treatment is being prescribed or shouldn’t be prescribed,” he stressed.
“It has to be transparent.” He stressed the financial discipline required: “We have to phase that investment in a manner that the benefits come through in a timely fashion.”
Taneja closed the session by outlining near-term priorities. “The immediate task in front of us for the next six months is to have a sort of agentic AI board in various ICSs, identify two, three enabled scenarios where this agentic AI can definitely make a difference,” he shared.
“We have started the FDP implementation. We need to engage with clinicians and design these agentic AI processes.”
The session ended with a wide-ranging audience Q&A. Asked about the biggest roadblock to AI adoption, Taneja said: “The biggest immediate challenge that we face is integrating the data that is sitting in multiple organizations, multiple places, multiple formats… without having access to proper data, any sort of agent will fail.”
On tackling it, he added: “FHIR standards… is the first step. The second stage would be to really get these agents in place… working with what we have currently.”
Taneja was also asked whether full interoperability with private providers, including dental, would affect timelines. “Interoperability would definitely slow down the adaptation of multiple use cases,” he explained. “But some use cases currently can be implemented right now even without perfect interoperability.”
Another question sought advice on communicating complex architectures and agentic concepts to NHS leadership. Taneja responded: “Explaining, having a clear vision is just the starter. Main thing is to get the ground level use cases going… we will write business cases saying that if we implement this use case, this is a potential saving or benefit to the patient.”
Taneja closed simply: “That’s all we do.”
Why not become a QA Financial subscriber?
It’s entirely FREE!
* Receive our weekly newsletter every Wednesday * Get priority invitations to our Forum events *
REGULATION & COMPLIANCE
Looking for more news on regulations and compliance requirements driving developments in software quality engineering at financial firms? Visit our dedicated Regulation & Compliance page here.
READ MORE
- Inside JPMorgan’s $18bn QA push with OmniAI reshaping testing
- As AI takes hold, insurance firms face a new testing mandate
- K2view’s Amitai Richman calls out the ‘real bottleneck’ in healthcare and insurance
- AI in QA: how flexible testing is redefining assurance for financial firms
- Explainer: Why site reliability engineering is gaining momentum in banking
WATCH NOW
