What is the Federated Data Platform?

Most of the conversation about FDP is about whether it should exist, or who should supply it. Almost nobody is talking about what it can actually do. That is a problem, because what is coming next is significantly more powerful than what has been delivered so far. The organisations that understand that now will be years ahead of those that don't.

I spent three and a half years leading the engineering team that built FDP at NHS England, and nearly 20 years before that building data platforms inside Trusts. I have seen both sides: the national architecture and the local reality. What I am about to describe is not theoretical. The foundations are already in place. Most people just haven't been told about them yet.

The Elephant in the room

Henry Mintzberg's Strategy Safari observed that people argued endlessly about what "strategy" meant because they were each standing too close to the elephant, describing a trunk, a tail, or a leg, and insisting it was the whole animal. FDP has the same problem. To researchers it looks like OpenSAFELY and they wonder why the NHS needs another research platform. To data engineers it looks like Databricks or Snowflake and they wonder why the NHS is paying a premium for something they could build themselves. To analysts it is a national reporting layer across Trusts and they judge it solely on whether the dashboards are better than Power BI. To operational staff it is the thing that replaced their whiteboard with a screen. To politicians it is a Palantir contract.

They are all describing real parts of the platform. None of them is describing the whole thing.

This matters because much of the criticism of FDP is based on examining one part of the elephant and concluding the whole animal is a failure. The BMJ recently published a critique of the benefits methodology for one FDP product at one Trust. That is legitimate scrutiny of one component. But using it to dismiss the entire platform is like inspecting the tail and concluding the elephant cannot walk. The theatre scheduling tool is one of six initial products. The platform itself is capable of far more than any individual product. The debate about what FDP is will not be settled by examining any single use case, because FDP is not a single use case. It is infrastructure.

There are calls for a debate about what FDP actually is. This post is my attempt to answer that question, based on three and a half years of building it from the inside.

The calculator app on a smartphone

Right now, Trusts that have adopted FDP have access to a handful of nationally developed products. Waiting list validation. Discharge planning. Theatre scheduling. Cancer pathway tracking. These are useful, and they are delivering measurable results. 110,000 additional patients have had operations. Nearly 300,000 have been discharged faster. Cancer diagnosis times have improved.

But these products are the calculator app on a device that could run anything you can imagine.

The programme deliberately kept things constrained in the early phases. There were good reasons for that. FDP was politically sensitive from day one, and the team wanted to bank real benefits before talking about the bigger vision. That caution made sense at the time but it has had a cost: very few people, even those quite close to the programme, understand what the platform is actually capable of.

Five things are about to change that.

1. Trusts are starting to replace their data warehouses

Almost all Trusts run a local data warehouse. Many data warehouses run on Microsoft SQL Server, Databricks, or something similar. Trusts pay for licensing. They pay staff to maintain it. They pay again to extract, transform, and submit national data returns manually, a process that eats up analyst time every month and rarely produces an output that has high fidelity to what happened in the clinical environment.

Trusts are already exercising the option to use FDP as their core data infrastructure instead. The Trust still controls its own data. It still decides who can see what. But instead of maintaining its own bespoke infrastructure, it runs on a centrally funded, cloud-based platform. National submissions happen automatically through the platform rather than through manual extracts and uploads, freeing up Trust data teams to focus on higher value add tasks.

FDP is centrally funded and is almost certain to stay that way beyond the seven year contract. The financial model would collapse overnight if Trusts were expected to fund this locally, which is why central funding is almost certain to continue. But NHSE has not said this publicly, and the silence is creating unnecessary uncertainty. This is an area where politicians could be asking useful questions.

The financial benefits for a Trust paying £100k a year or more in Microsoft licensing alone, plus the staff costs to keep it all running, this is a significant financial saving. It also moves the Trust from on-premise to cloud without the usual cost and complexity of a cloud migration programme. The infrastructure is already there, centrally funded, waiting to be used.

When I worked in Trusts, we did invest in the data warehouse, but under very limited resourcing and with no technical direction from the centre whatsoever. NHS Digital did its bit but was limited to ensuring national data collection happened. The Trust data warehouse sat in the corner, held together by a handful of people who understood it, and when they left the organisation the knowledge walked out the door with them. Its development path was completely reliant on the small local team’s level of expertise and whether the Trust leadership with its largely clinical background would support their vision. FDP changes that by providing a standard, supported, nationally maintained platform underneath.

This makes the data workforce much easier to develop: by getting all engineering staff onto the same technology it allows for the first time a consistent engineering training programme for all of the NHS. My conservative estimate for the number of NHS-wide engineering posts would be around 1,500 to 2,000 specialist staff. NHS England has just committed to supporting data as a profession, and the job has just got easier because it doesn’t have to support learning across dozens of different technologies. Right now it’s hard to find new recruits to a data team with the right skillsets but as more Trusts adopt FDP as their warehouse this will get easier and a large workforce will emerge.

It also allows for a consistent development roadmap for this part of the NHS data infrastructure. No longer will each Trust’s data estate worm its way near-blindly through the rapidly changing technology landscape – a process being accelerated due to the advent of AI and an increasing willingness to invest in this space. Instead it will be centrally guided and locally informed in a far more collaborative and technologically aware manner, reducing the current acceleration of the fragmentation of the NHS’s data estate.

2. Trusts are replacing shadow IT with robust applications

But FDP is much more than a data warehouse. Traditional data warehouses sat at a distance from the operational data systems. FDP does not. It operates across both primary and secondary uses of data. This is what makes it so powerful in comparison to other cloud data warehouses which perpetuate the old pattern of secondary data use. In bridging this gap, FDP brings the potential to reduce data quality issues to a huge extent. A primary reason for poor data quality in secondary use cases is that the clinician got no benefit from entering that data. When they are using it day to day, and getting something back, the quality is near perfect.

Every Trust has shadow IT. Spreadsheets tracking patient lists. Whiteboards in discharge lounges. Emails coordinating theatre schedules. Printed lists that get updated with a biro during ward rounds. I did an audit at one Trust and found over 1,000 non-approved data sources including spreadsheets and paper records.

These are not just inefficient. They are patient safety and information governance risks hiding in plain sight. A spreadsheet can be out of date the moment it is saved. A whiteboard can be misread. An email can be missed. Any visitor walking through a ward can see the whiteboard in the nurse station. There is no audit trail, no access control, and no way of knowing whether the information is current. The information is lost the moment the whiteboard eraser is used.

FDP gives Trusts the tools to build proper applications that replace this shadow IT. And the Build with FDP event proved that you don't need to be a software engineer to do it. In October 2025, 120 NHS staff spent two days building digital tools on the platform. The winning teams included paramedics and operational staff who had never written code. The platform's AI-assisted development tools allowed them to build working prototypes for real frontline problems in 48 hours. The next event is in Leeds in May 2026. The best solutions are being sponsored by NHS England for further development and national roll out.

This is what the platform looks like when you stop arguing about the supplier and start using it.

3. Connect data across care settings

This is the capability that could change the most for patients, and almost nobody is talking about it. Some folks still think FDP is all about business intelligence and data for secondary use. They are wrong, it is an operational platform as well.

A mental health patient arrives in A&E. The acute Trust runs Epic. The mental health liaison team uses RiO. These systems do not talk to each other. The A&E clinician cannot see the patient's mental health history. The liaison team cannot see what has been done in A&E. The patient waits. Sometimes for hours. Sometimes in entirely the wrong setting.

FDP can bridge that gap because both organisations can be on the same platform, both using a niche app, that they could co-develop, that not only surfaces the data but allows them to add to it. Data about the patient's journey can flow between the acute Trust and the mental health Trust without anyone having to reformat, re-enter, or manually transfer information between systems.

This is not just about mental health. It applies to any patient whose care crosses organisational boundaries: frail elderly patients moving between hospital and community services, cancer patients being referred between a district general and a specialist centre, children with complex needs being seen by multiple teams across multiple organisations.

The NHS has been talking about integrated care for years. FDP is the first piece of infrastructure that actually makes it possible at scale, because it gives organisations a common data platform that works across care settings rather than within them.

4. Share and scale what works

When every Trust is running on the same platform with the same data model, something changes fundamentally.

A Trust in Dorset builds a dashboard that helps its clinical team manage patient flow. Under the old model, a Trust in Newcastle that wanted the same thing would have to build it from scratch because the two organisations run on completely different technology. It would be difficult for them to even become aware of Dorset’s work. On FDP, Newcastle can find and adopt Dorset's tool without rewriting a single line of code.

Staff can move between organisations and be productive on day one. Right now, a data engineer moving from one Trust to another has to learn an entirely new set of systems, tools, and data structures. On FDP, they already know the platform. Not only is this a productivity improvement, it allows deep and widespread knowledge to develop, and prevents rework.

Technology companies can build products that integrate with a standard data model and platform across the entire NHS, rather than doing bespoke integration work at every site. Build once, deploy everywhere. That dramatically reduces cost and time to market. The Solutions Exchange, which is essentially an app store for FDP, will create a marketplace where third parties can publish applications for any Trust to find and adopt. NHS England has already started engaging with suppliers. The framework is being developed for 2026/27.

And this standardisation survives regardless of who supplies the platform technology. Even if Palantir were replaced tomorrow, the data model and the architecture would carry forward. That is the thing most people miss when they focus exclusively on the supplier.

5. AI that actually understands NHS data

This is the one that could change everything, and it is built on something most people outside the programme have never heard of: the Ontology.

The Foundry Ontology, a proprietary use of the word traditionally used in data science to mean something different, is not just a related set of objects that store rows of data. It is a structured representation of how NHS data relates to itself containing all relevant labels and metadata. It couples the data directly with the NHS Canonical Data Model. It knows what a Trust is. It knows what a waiting list is. It knows what an ASD referral means in clinical terms.

It connects data that have traditionally been held entirely separately to clinical data such as financial and workforce data, making the overall data source far more useful for combatting NHS inefficiency. It understands how these concepts connect to each other. Moreover, the ontology has the capability not just to surface data but to support edits to the data and this is the basis of true FDP applications not just dashboards.

That matters because it gives NHS staff the means to alter data surfaced in an FDP app without having to click back into many different source systems across different organisations. But much more than that it offers AI agents something that no other health data system in the country can offer: not just the data, but the meaning behind it.

NHS England is trialling AI tools that allow users to ask questions of the data in plain English. A manager could ask "which Trust has the longest wait for ASD assessment in children?" and the system will run the analysis, drawing on the ontology to understand what that question means, which data to query, and how to interpret the results. No analyst required. No SQL query. No waiting three weeks for someone to extract the data and build a report.

This has the potential to radically reduce the analytical burden on Trusts and ICBs. The analytical workforce is one of the most stretched in the NHS. If AI can handle the routine queries, like performance monitoring, benchmarking, and standard reporting, it frees up those teams to do the complex work that actually requires human judgement: investigating why something is happening, not just confirming what is happening.

It also addresses a criticism that I think is fair: the reporting and analytics tools currently available on FDP are not as strong as those some Trusts already use, such as Power BI. For example, Fusion Sheets has limits that even Excel can beat. Some of the current analytics tools have limitations that experienced analysts find frustrating. But the direction of travel is that you will not need a traditional reporting tool for many questions, because you will simply ask the platform directly. The current analytics limitations are a transitional issue, not a permanent one. For those that continue to use Power BI, FDP allows that too.

So what does this mean for you?

If you lead a Trust: there are huge savings to be found at the interfaces between clinical processes that have not been looked at or optimised before because the data was siloed by organisation or service boundaries. The organisations getting ahead right now are the ones treating FDP as an operational capability, not an IT compliance exercise. They are replacing costly local infrastructure, eliminating shadow IT, connecting data across care settings, and building bespoke tools for their clinical teams. If you are waiting to see what happens with the political debate before making a decision, you are falling behind Trusts that are not waiting.

I spent 20 years on the frontline wishing I had something like this. I watched clinicians queue up on Fridays to enter data from paper notes. I watched managers walk around wards with printed spreadsheets and biros. I watched organisations spend millions building data systems that could not talk to the Trust next door. FDP is the first time anyone has built the infrastructure to fix that at national scale. The Trusts that use this window to get their foundations right will be years ahead as the platform matures.

If you run a technology company: the Solutions Exchange is about to create a market that has never existed. Companies that understand FDP architecture and the national data model now will be first to market when it opens. Companies that are sitting on the sidelines because of the political noise will arrive late to a market that has already been claimed by their competitors.

I have spoken to technology companies at conferences who told me they are holding off because they are not sure FDP will survive. I understand the caution. But the platform is rated green by NISTA, 135+ Trusts are live, and there is no credible alternative. The risk is not that FDP disappears. The risk is that you are not ready when it takes off.

What comes next

I am publishing a series of posts over the coming weeks going deeper into each of these areas. I am also developing the FDP Playbook, a practical guide for Trust leaders and technology companies that want to understand the platform without wading through programme jargon or political noise. It will be available on this site shortly.

If you want to understand what FDP can do for your organisation, or if you are a technology company trying to work out what the NHS data landscape looks like, I can help.

Tom Bartlett

Founder, Bartlett Data Ltd

tom.bartlett.data@outlook.com