Healthcare digitisation and software modernisation
Healthcare digitisation turns paper, phone calls and old software into connected systems that staff can rely on, in small steps that never stop care.
We do it for private hospital groups, clinic groups, NHS trusts, community services and the clinical operations teams of health payers. AI does the reading, sorting and checking at each step, and a named person at your organisation approves its work.
Last updated
What do we digitise?
The paper, the phone calls, the systems that do not talk to each other and the old software that holds your rules. Each piece of work names the person who approves it.
-
Paper records and forms to structured data
Scanned notes, consent forms and questionnaires are read by a document intake agent, which pulls out each field, checks it against rules you set, and flags anything it is unsure of instead of guessing.
Who approves: A member of your records team reviews every flagged field, and checks a sample of the rest, before the data is used.
About the document intake agent -
Referral and letter workflows
Incoming referrals and clinic letters are sorted, matched to the right patient and routed to the right team, with missing information requested before a referral waits in a queue.
Who approves: A named administrator confirms each match and route, and a clinician makes every triage and clinical decision.
-
Phone and booking
Prime Assist, our AI voice infrastructure for healthcare, answers routine calls, books and moves appointments in your own diary and passes anything else to your team. It is for private hospitals and clinics; it is not yet used in the NHS.
Who approves: Your practice or bookings manager sets what it may handle, and every call it cannot finish goes to a person.
About Prime Assist -
Disconnected systems, connected
We connect patient administration, booking, records and finance systems through HL7 FHIR, using the UK Core profiles where they fit, and through each system's own published interfaces, so staff stop typing the same details twice.
Who approves: Each system's owner and your information governance lead approve every data flow before it goes live.
-
Old software, modernised in stages
We review the code first, as we do in our technical due diligence work. Then we write down the business rules the old system holds and write tests that prove its current behaviour, before anything is replaced, one part at a time.
Who approves: Your system owner signs off the documented rules and the test results before each part is switched over.
About our code review work
Where does AI help, and who approves it?
At every step, AI does the repetitive reading, matching and checking, and a named person approves the result. Nothing it produces changes a record or reaches a patient without that person.
How our AI agents work| Step | What the AI does | Who approves |
|---|---|---|
| Reading documents | Extracts fields from scans and forms and scores how sure it is of each one. | Records officer |
| Sorting referrals and letters | Suggests the patient match, the team and what is missing. | Referral administrator, with clinicians deciding on care |
| Calls and bookings | Answers routine requests and books within the rules you set. | Bookings manager |
| Connecting systems | Maps fields between systems and flags records that do not agree. | System owner and information governance lead |
| Modernising software | Helps read old code, draft the business rules and generate tests. | Your system owner and our lead engineer |
See the agents for each sector: NHS and public sector and private healthcare.
Lean and fast
Small releases, each one useful on its own, so your teams see a change working before the next one starts.
- Small releases: one form, one referral route or one part of an old system at a time.
- Old and new side by side: the new way runs next to the old until your team is satisfied, so there is always a way back.
- Measured against today: before we change anything, we record how the work is done now, and each release is compared with that.
- Only what is needed: we keep the systems that work and change the ones that hold you back.
Who decides: Your service lead decides when each release goes live and when the old way is switched off.
Why digitise now?
National plans in the UK and the EU assume records and referrals are digital and connected. These are the dates, with the source for each.
-
By April 2028
AI triage in the NHS App
NHS England has said AI triage will be available in the NHS App for every user by April 2028. Services behind it need referral and booking data they can act on.
-
From 2028
A single patient record
The Health Bill proposes a single patient record that patients reach through the NHS App from 2028, subject to Parliament. Records still on paper or in closed systems cannot take part.
Source: GOV.UK: Health Bill single patient record fact sheet
-
From March 2029
European Health Data Space
Under Regulation (EU) 2025/327, patient summaries are due to be exchanged in a common format across the EU from March 2029, which matters to providers and payers with patients in EU countries.
Dates as they stood on 26 September 2026. Plans and dates change, so every project starts by checking where each one stands.
How is clinical safety handled?
Throughout, from the first assessment to every release. What we build is managed under DCB0129, and we support your team with DCB0160 for what you deploy.
Clinical safety for AI you deploy- DCB0129 for what we build: a hazard log and a clinical safety case for each release, kept up to date as it changes.
- DCB0160 support for what you deploy: the evidence your Clinical Safety Officer needs to assess and sign off each change in your setting.
- Clinical decisions stay with clinicians. AI prepares and sorts; it does not diagnose or decide on care.
Who decides: Your Clinical Safety Officer approves each change for use in your organisation.
How do we approach green AI?
By how we design each AI step. Our green AI work is in development, and this is what we do today.
Read about our green AI work- The smallest model that does the job: each task goes to the smallest model that meets the pass mark agreed with you.
- On the device where possible: routine requests are answered on the device with Prime Edge AI, which we are developing.
- The cloud only when needed: a request goes to a larger model in the cloud only when the task needs it.
- Measured with a published method: we are working on measuring energy per request with the Green Software Foundation's Software Carbon Intensity (SCI) for AI, so the method can be checked.
How a project runs
Four steps, starting with an AI readiness assessment of how the work is done today.
-
Assess
Map how the work is done today, where the data lives and which systems hold it, and agree who approves each change.
-
Plan
Choose the first small release, the measures to compare it with, and the clinical safety work it needs.
-
Build
Build, test and run it next to the old way, with your approvers checking the results.
-
Move over
Switch over when your approvers are satisfied, then start the next release.
Questions people ask
What is healthcare digitisation?
Moving the paper, phone and manual steps of running care into connected digital systems, so information is captured once, reaches the right person and can be checked. Done well, it is a series of small changes rather than one large switch.
Do you replace our patient record system?
Not unless you decide to. Most of our work connects and improves the systems you already run. Where old software has to go, we document its rules and test its behaviour first, then replace it in stages while the old and new run side by side.
Do you work with EMIS, SystmOne or other named systems?
We connect to systems through HL7 FHIR and each system's own published interfaces. Which connections are possible depends on each supplier's access terms, so we confirm them with you and the supplier at the start of each project.
Who is responsible for the AI's decisions?
Your people. Every AI step on this page has a named person at your organisation who approves its work, and clinical decisions always stay with clinicians. We build what we supply under DCB0129, and support your team with DCB0160.
How do you approach green AI?
We choose the smallest model that does each job, answer routine requests on the device where possible and send them to the cloud only when needed. We are working on measuring energy per request with a published method. This work is in development.