Dr Shankar Sridharan, NHS England's national clinical lead for AI, had a blunt message at NHS ConfedExpo in June. Trusts keep running pilots of ambient voice technology (AVT), he said, even though the evidence is already there. According to Digital Health, he called continued AVT pilots "nuts" and asked how much evidence is enough.

The money is following that message. NHS England is rolling out AI tools, including integrated ambient voice technology, backed by £10bn of technology funding over 3 years. If you're building a product for the NHS, NHS healthtech procurement is changing under your feet.

Finding a friendly trust and running a small pilot used to be a sensible way in. Now it gets you a polite conversation and not much else. Buyers want evidence, a route to scale, and proof that your product works for thousands of clinicians, not 20.

Below, we cover what's changing, what 1 national contract tells us (and what it doesn't), what evidence buyers now expect, and what you can build in from day 1.

How NHS healthtech procurement is moving from pilots to scale

NHS healthtech procurement is shifting from local trials to buying proven tools at scale. NHS England is rolling out AI tools, including integrated ambient voice technology, backed by £10bn of technology funding over 3 years. Senior leaders are urging trusts to stop repeating pilots of tools that already have evidence.

You can see the shift in how AVT is being bought. In the Midlands, NHS England procured AVT for 1,239 GP practices and around 70,000 clinicians across 15 acute and community trusts, according to Digital Health. Heidi was selected as the preferred supplier. That's a regional decision covering a huge number of users, not a trust by trust experiment.

Nationally, NHS England launched a supplier registry for AVT in January 2026. NHS England Midlands reports that it lists 23 self-certified suppliers. NHS England doesn't endorse any supplier on the registry and only runs preliminary completion checks, and individual NHS bodies still carry out their own procurement.

There's also a national framework. NHS Shared Business Services set up a £150m framework for AVT. It runs from 1st October 2025 to 30th September 2027, with an option to extend by up to 2 years, as reported by Digital Health.

Why the NHS wants scale

The economics help explain the push. Economic modelling that accompanied a Great Ormond Street Hospital led trial looked at what happens if AI scribing reaches all 11,055 A&E clinicians in England. It estimated 9,259 extra A&E consultations each day, £176 million saved in documentation time and a further £658 million of capacity unlocked each year, as Digital Health reported. Numbers like that help explain the appetite for scale.

Funding doesn't guarantee smooth buying, though. After the Autumn Budget in November 2025, concerns were raised that capital funding could be spent on hardware and slow the rollout of tools like AVT, Digital Health reported. Money is moving, but where it lands still matters.

What pilots are still good for

Not every leader is calling for an end to piloting. The push to stop repeating pilots is about proven tools like AVT, where the evidence already exists. If your product is newer, you'll still need to build that evidence yourself.

The NHS AI rollout is most advanced in ambient voice, but the direction applies to anyone building in health. A pilot should be a step on a route to scale.

Before you start, agree what success looks like and how you'll measure it. Run it somewhere that reflects real conditions, with your product connected to the systems clinicians already use. Plan what happens if it works.

Are national single platform contracts the new default in NHS healthtech procurement?

In some categories, yes, and in others, not yet. The clearest recent example sits in diabetes care, and it comes with a long research history.

University Hospitals of Leicester NHS Trust has been awarded a 5 year contract to deliver NHS England's Healthy Living service. It gives adults with diabetes across England digital structured education and support to manage their condition. The service runs on MyDESMOND, a platform developed by the Leicester Diabetes Centre, and is due to launch in autumn 2026.

DESMOND grew out of research led in Leicester more than 20 years ago. The new contract adds a programme for adults with type 1 diabetes for the first time. Promatica Digital provides technical delivery, including integration with primary care systems.

Nelson Advisors, which advises on healthtech deals, read this award as a sign of things to come. In its UK HealthTech Pulse for 15 to 18 August 2026, it said national, single platform contracts like this are becoming the default route to scale in condition management. In its view, that's good news for platforms with an evidence base and harder territory for point solutions.

How far does that view stretch?

It's a useful signal, but it's 1 adviser's view built on 1 contract, so we'd treat "default" with care. AVT, for example, runs through a supplier registry, a national framework and regional deals, rather than a single national award.

Single supplier deals also carry their own risk. Nelson Advisors' analysis of the NHS Federated Data Platform notes that public sector technology leaders worry about vendor lock-in and high switching costs when 1 prime contractor holds everything.

Look at what these examples reward. The MyDESMOND award went to a programme with more than 20 years of research behind it. It's delivered by a partnership between a trust, a research centre and a technical delivery company.

If you're a point solution, that doesn't rule you out. It does mean you should plan how you plug into something bigger.

Why a working prototype isn't enough: the healthtech clinical evidence buyers expect

A working prototype shows that your product runs. Buyers now also want healthtech clinical evidence that it works, and proof it will hold up at national volumes.

The bar is easiest to see in the AI scribe trial led by Great Ormond Street Hospital. It evaluated more than 17,000 patient encounters across 9 NHS sites in London, between June 2024 and February 2025. Direct patient interaction time rose by 23.5%, and overall appointment length fell by 8.2%. In A&E, 13.4% more patients were seen per shift.

Those findings went on to inform NHS England's guidance on AI scribing and the NHS T.E.S.T. Framework, a national model for evaluating AI technologies. That's the kind of evidence behind the argument that further pilots add little.

The 2 frameworks behind most buying decisions

The NICE evidence standards framework (ESF) sets out the clinical and cost effectiveness evidence a digital health technology needs. What counts as enough depends on what the product does and the risk it carries. Meanwhile, the NHS Digital Technology Assessment Criteria (DTAC) sets minimum standards for digital health products, covering clinical safety, data protection, technical assurance, interoperability, and usability and accessibility.

NHS guidance says buyers should use both. Law firm Burges Salmon notes that a refreshed DTAC form replaced the previous version from 6 April 2026. Some categories, including AVT, also have extra standards on top.

The questions you'll be asked

Before anyone signs a regional or national contract, expect questions like these:

• Where is your clinical evidence, and which outcomes did you measure?
• How does it work for thousands of clinicians, not 20?
• How does it connect to the electronic patient record and GP systems already in use?
• Does it meet DTAC, and any extra standards for your category?
• What happens when it needs to cover a whole region, or the whole country?

‍

If you only start answering those once the product is built, you're already behind. Evidence and scale are easier to plan in than to bolt on.

How to start scaling healthtech in the NHS from day 1

Healthtech founders need a clinical evidence plan and a scaling story from day 1, not just a working prototype. That doesn't mean a multi year research programme before you write a line of code. It means treating evidence and scale as product requirements. Here's where we'd start.

Design for evidence

Pick your main outcome before you onboard your first user. Choose something a buyer already tracks, such as time saved per clinician, appointment length, waiting times or a clinical measure like HbA1c in diabetes care.

Then build the data capture into the product, so you can report on that outcome without a manual exercise later. Check which ESF tier your product falls into early, because it tells you what level of evidence to plan for. And agree an evaluation approach with an academic or clinical partner while the product is still cheap to change.

Build for scale, not just for the pilot

Design for the numbers you want to reach, not the ones you start with. The Midlands AVT procurement covers around 70,000 clinicians. Ask what breaks at 1,000 users, then at 10,000 and beyond.

Plan integration with the electronic patient record and GP systems from the start. Treat the DTAC areas of clinical safety, data protection and interoperability as design requirements, not a form to complete at the end. Keep your documentation up to date as you build, so a buyer's questions don't send you back to the drawing board.

Think platform, not feature

Be honest about whether you're a point solution. If you are, work out which larger platform or pathway you plug into, and make that easy with open interfaces.

MyDESMOND brings education, health tracking, peer support and activity challenges together in 1 service, and it was delivered with a technical partner that specialises in primary care integration. If you can't explain where you fit in a pathway, buyers will find it harder to say yes. Not every product should become a platform. Some are better as a well connected part of someone else's.

Get clinical credibility in the room early

Bring a clinical lead into product decisions before you build, not after launch. They'll spot workflow problems, help shape the outcomes you measure, and tell you what a clinical reviewer will question. Doing that work after the build can mean reworking features you've already shipped.

Why clinical expertise belongs in the room before you build

We've seen the pilot stage up close. We were among the founding collaborators on The Wayback, a VR app that takes people living with dementia back to happy historical moments. It began as a crowdfunded pilot, and it's now being rebooted for care homes, including a partnership with Belmont Healthcare. You can read more in our The Wayback case study.

The journey from pilot to partnership is the one NHS buyers now want founders to plan for. That's why Dr Jarama Clucas, known as Jazz, has joined us as our Scientific and Medical Advisor.

Jazz spent more than 7 years as a postdoctoral researcher at the Institute of Cancer Research in London, investigating immunogenic cell death in breast cancer. Before that, Jazz was a project research scientist at the Francis Crick Institute, and has also tutored medical and biomedical students.

Jazz's role is to make sure our digital products and services, especially in health tech, are grounded in solid evidence from discovery through to launch and beyond. The aim is for that thinking to be there from the first conversation, when it has the biggest effect on your decisions. If you're planning a health product, see how we approach discovery.

What to do next

NHS healthtech procurement now rewards proven tools with a clear route to scale. A polished demo matters less than evidence, integration and a clear place in the care pathway. And the earlier you plan for those, the less you'll have to rework.

If you're building in health and want to pressure test your evidence and scaling plan before you build, we'd be glad to talk it through.

Frequently asked questions

What is ambient voice technology (AVT)?

AVT records the conversation between a clinician and a patient. Combined with generative AI, it turns that speech into structured medical documentation such as notes and letters. The clinician checks and edits the draft before it's saved to the patient record, according to NHS England.

How does the NHS buy ambient voice technology?

There's no single route. NHS England runs a self-certified supplier registry, NHS Shared Business Services has a £150m framework, and regions such as the Midlands have run their own procurement. NHS England says individual NHS bodies still carry out procurement under their own governance.

What is DTAC and do I need it?

The Digital Technology Assessment Criteria is NHS England's assessment framework for digital health technology products. It covers clinical safety, data protection, technical assurance, interoperability, and usability and accessibility. If you're selling software that handles health and care data into the NHS, expect to complete it.

What evidence do NHS buyers expect from a digital health product?

It depends on what your product does and the risk it carries. NICE's evidence standards framework sets out the levels of evidence of clinical and cost effectiveness to aim for, and DTAC covers the minimum requirements for use in the NHS. Larger studies, like the Great Ormond Street Hospital trial of 17,000 encounters, show what strong evidence can look like.

Do NHS pilots lead to national contracts?

Not automatically. NHS leaders are pushing to stop repeating pilots of tools that already have evidence. A pilot only helps if it adds to that evidence and shows a route to scale. The MyDESMOND contract went to a programme backed by more than 20 years of research.

Is a point solution still viable in the NHS?

It can be, but plan for how you fit into a wider pathway. Nelson Advisors, an adviser on healthtech deals, sees national platform contracts as harder territory for point solutions. Strong evidence and clean integration with the systems clinicians already use give you the best chance.