Your browser doesn’t support HTML5 audio
Erika Bannerman, Managing Director of NHS Shared Business Services, was speaking at Oracle's AI World Tour London last week, where she said that the NHS is not a single organization, but:
Systems within systems, with multiple hundreds of organizations, software platforms, and entities that are all dependent on each other.
This is typically why transformation programmes across the NHS have failed, stalled or delivered mixed results. The demands or opportunities of a ‘national entity’ don’t always align well with the restraints or requirements of local needs. Managing the financial infrastructure that sits underneath all of that - the cash flows, the supplier payments, the procurement pipelines - is an enormous undertaking. NHS SBS, however, does it on behalf of the health service, and the scale of the NHS means that it is processing up to £355 billion in transactions annually.
The organization recently went live with a new national finance and procurement platform built on Oracle Fusion Cloud Applications. The platform launched on 1 October 2025, bringing 48 NHS organizations onto a single, cloud-based finance system. The work that preceded it took years, with around 30,000 hours of workshops, a 473-item requirements traceability matrix, and a squad-based approach that brought together developers, users, subject matter experts, and customers.
And yet, sitting at an AI event in 2026, Bannerman acknowledged the timing was both fortuitous and interesting. Bannerman said:
We may have been a year early. We went live with 48 organizations on something genuinely transformational, and we've been going through all the usual stages of hypercare. We now have a stable, secure, effective platform that is working. We've still got to realize so many of the benefits that will come from it.
The AI capabilities that Oracle and others are now bringing to market at pace did not fully exist when NHS SBS was designing the platform four years ago. The organization built something agile enough to adopt them as they arrived - and NHS SBS believes that the foundation is now in place to properly take advantage of extensive AI tooling.
The scale of the challenge
It is worth reiterating what NHS SBS actually manages, because the numbers are at a scale that would cause pause amongst even the largest, most sophisticated enterprises before considering any platform changes. The organization processes 7.1 million invoices annually, has recovered £7.4 billion in debt, and handles up to £355 billion in NHS transactions per year - a figure that, as Bannerman noted, exceeds the NHS budget itself, because it captures the full flow of cash through the system rather than net spend.
On day one of the go-live, the team needed to process £19 billion. Bannerman said:
We needed to be sure we had the security, the governance, the controls - that we could ensure the money was getting to the front line, to pay NHS employees and suppliers. It was critical that it worked.
Early feedback from front-line teams was that the system was intuitive, required minimal training, and was already cutting task times in half. NHS SBS has also reported more than half a million pounds in cost savings from migrating to Oracle Exadata Database Service on Oracle Cloud Infrastructure, alongside 30 percent cash flow savings in the platform's first year.
The AI opportunity
As one would expect, the shift to a cloud-based, evergreen platform matters also creates the conditions for AI adoption. This is the part of the story that Bannerman is most animated about, even though she recognizes that it’s still early days.
The NHS has 1.6 million employees. Corporate functions - finance, procurement, HR - sit across hundreds of sovereign entities, each making its own decisions about technology investment. Bannerman's argument is that a national shared services platform changes that dynamic, in that it provides the standardized data infrastructure that AI requires, at a scale no individual trust or integrated care system could replicate.
She identified four prerequisites for making AI work in this environment: end-to-end process orchestration rather than task automation in isolation; standardization; ethical governance; and a clear data strategy. On the last point, Bannerman said:
We have a lot of unstructured data within the NHS. What's super exciting now is what we're going to be able to do with that, and the real difference it will make to the front line.
The embedded AI capabilities within Oracle Fusion - spanning ERP, EPM, and the broader application suite - mean NHS SBS does not need to build AI on top of its platform. It is already there, and each regular release brings more. For a health system that has historically fallen behind on software updates simply because the cost and complexity of upgrading was prohibitive, that holds potential for value to be captured regularly (if that change management challenge can follow).
The £3 billion question
The headline figure from the keynote was a potential £3 billion in annual savings for the NHS if the whole system adopted the NHS SBS platform - an unlikely scenario, but one that highlights the opportunity if the NHS was to standardize on back office systems. Bannerman said:
That's a lot of nurses. That's a lot of hip operations. There's a lot to go after there.
However, the savings are contingent on adoption, and adoption is where things get complicated. Corporate shared services, she noted, are "always at the bottom of the investment pile." The challenge is not the technology. It is affordability, change management, and the difficulty of asking organizations already under significant pressure to prioritize back-office transformation.
Bannerman said:
Everybody wants to change, transform, and adopt technology, but the budgets and the finances aren't just for the technology itself, they're for the change management required to make it happen.
This is why the self-funding joint venture model holds potential. NHS SBS reinvests income back into the system to help organizations fund the transition, including the change management costs that would otherwise make adoption unaffordable. Compete, win contracts, generate income, reinvest, deliver impact. It is, as Bannerman described it, a virtuous circle.
What comes next
Bannerman said that her team is now exploring virtual agents. The question they are asking internally is, regarding use cases, is:
Rather than inducting another new starter, what if we were to induct a virtual agent? How would you manage that virtual agent? How would you scale that?
It is an early-stage conversation, which Bannerman was honest about. However, the 10-year strategic plan also extends well beyond finance into procurement-led transformation: understanding demand, managing resilient supply chains, giving suppliers visibility of pipelines, and using data to drive down costs.