Clinically Intelligent is written by Juwon Akinyande in a personal capacity and is not affiliated with, endorsed by, or representative of Barts Health NHS Trust or any other NHS organisation. Content is for general educational purposes only. It does not constitute clinical, legal, or information governance advice. Before applying any guidance to your own practice, consult your Trust information governance lead, your Caldicott Guardian, your line manager, and your professional body.
The biggest NHS AI rollout in history just got annouced.
I read the announcement twice. The headline is interesting. 505,000 NHS staff. Microsoft 365 Copilot. The largest AI rollout in NHS history. 43 minutes saved per person per day. Five working weeks per person per year.
Then I read the use cases.
Ward clerks helping with patient discharge processes and bed management. Medical secretaries drafting patient letters and meeting minutes. HR, finance, and procurement functions. Board papers and briefings.
The one clinical use case named is clinical administration — assisting clinicians in drafting letters and registrar training. That is worth mentioning. It is also worth mentioning that drafting letters is the same task a medical secretary is doing two bullet points down the same list.
The announcement comes across as a revolution for patient care. The closest it gets to clinical work is better paperwork.
That gap between the announcement and the detail is the thing worth paying attention to.
What is being rolled out.
Microsoft 365 Copilot is the paid version of Microsoft's AI assistant. It sits inside the tools every NHS clinician already uses. Word, Excel, Outlook, Teams, PowerPoint.
Most NHS staff have already had access to a free version of Copilot Chat for some time. You sign in to copilot.microsoft.com with your NHS account and you have a chat tool comparable to ChatGPT or Claude. That free tier has been available for months. Almost nobody uses it.
The paid version that 505,000 staff are now getting is different. It is not a separate chat window. It is AI built directly inside the apps you already open. Copilot inside Outlook drafting replies using your email history. Copilot inside Word drafting documents using context from your SharePoint. Copilot inside Teams summarising meetings and generating action items. Copilot inside Excel building formulas from natural language and analysing data.
That distinction matters. The free tier is the AI you go to when you remember to. The paid tier is the AI that is already there when you open the document. Tools that live inside your existing work get used. Tools that require a separate tab usually do not.
What the announcement actually says.
The use cases NHS England published cover the operational side of healthcare rather than clinical work. Clinical administration. Patient discharge processes. Medical secretary work. Ward and bed management. Corporate functions across HR, finance, and procurement.
These are the tasks that fill the spaces between care. The announcement is not about AI replacing clinical judgement. It is about AI reducing the time spent on the work that surrounds clinical judgement. That is a legitimate and worthwhile goal.
But the framing of the announcement and the content of the announcement are two different things. The headline says this is groundbreaking for patients and clinicians. The detail says here is a Copilot licence for the people who run the admin. That is not a criticism of the ambition. It is an honest reading of what has actually been announced versus what has been promised.
The 43 minute figure is not 43 minutes saved on patient assessment. It is 43 minutes saved on documentation, scheduling, and correspondence. Whether those 43 minutes become time with patients or get absorbed by other backlogged work is a question the announcement does not answer.
How licences are being distributed.
Each NHS Trust will receive a central allocation of licences based on organisational headcount, typically starting at around 2,000 licences per Trust. 200,000 users are expected to be scaled up within the first six months. The full 505,000 rollout is expected by October 2026.
2,000 licences per Trust is a meaningful allocation but it is not universal coverage. A large acute Trust has thousands of staff. The initial licences will be distributed somewhere but not to everyone.
Who gets the licences in your Trust is a procurement decision that has not been made yet for most Trusts. The honest reading of the announcement is that whether you get a licence depends on choices your Trust digital lead is making in the next six months.
What is missing from the announcement.
The biggest gap is training. NHS England mentions a 12 month onboarding programme with extensive training and adoption support. The detail of what that looks like in practice has not been published.
This matters because most NHS staff have already had access to Copilot Chat for months and have not touched it once. Giving more staff access to a tool they are not using does not equal more people using it. Access and adoption are two different problems. The announcement solves the first one. The second one is harder and nobody has explained how they are going to solve it.
Tools deployed without training get used badly, get abandoned, or get used in ways that create governance risks. Rolling out 505,000 licences without a clear answer on what the training looks like is the part of this announcement worth watching most carefully.
The other gaps worth naming. Which clinical staff groups specifically get the initial licences and whether AHPs are included. What the clinical safety case looks like for Copilot operating inside documents that may contain patient information. How the 43 minute time saving will be measured at Trust level given how much local measurement infrastructure varies.
These are not criticisms of the announcement. Announcements are necessarily high level. They are the questions that determine whether this becomes operational reality in your service or stays as a headline.
What to do this week.
Three practical actions.
One. Find out whether your Trust is part of the initial scale up to 200,000 users in the first six months. Your Trust digital lead or AHP lead should know. If they do not, that is itself useful information about how integrated AHPs are in the local rollout planning.
Two. Use the free Copilot Chat today. Sign in at copilot.microsoft.com with your NHS account. Try it on a clinical research question, an email drafting task, and a document summarisation. Ten minutes will tell you whether the chat tool fits your workflow. The paid version being rolled out is the same underlying technology embedded across more places. If the free tool fits, the paid integration will fit more.
Three. Run the Safety Check from Issue 07 against Copilot Chat as you test it. The framework applies whether the tool is consumer AI or enterprise AI. Composite scenarios. No patient identifiable information. The data handling under your NHS enterprise account is more protective than consumer AI but the underlying discipline does not change.
Information Governance.
The data handling under your NHS Microsoft account is more protective than consumer AI tools. Microsoft has committed that prompts and responses in Copilot under enterprise accounts are not used to train Microsoft's AI models. The clinical safety case for the enterprise deployment will be assessed against DCB0129 and DCB0160 standards as part of the procurement.
That does not remove the need for the Safety Check framework. The institutional procurement covers the tool. The clinician using the tool is still responsible for what they put into it. Composite scenarios. No patient identifiable information. The framework from Issue 07 applies.
The full Safety Check is at clinicallyintelligent.com.
Opinion.
A licence is not a plan.
The use cases in this announcement are admin tasks. Every single one. That tells you something about where the understanding of AI in healthcare currently sits at a national level. The people writing the announcement know how to deploy a licence. They have not yet figured out how a Band 6 OT uses AI on a Tuesday afternoon with a full caseload. That is not an insult to anyone's title or intentions. It is just an honest read of what was published.
The honest pattern of NHS digital change is that the headline arrives, the detail follows, the delivery follows, and the actual change to clinical practice follows years later if at all. The announcement is necessary but not sufficient. A rollout is not an implementation. A licence is not a plan.
The clinicians whose work would most change from AI integration across Microsoft tools are usually not the people making the procurement decisions. The decisions are technical and financial. The case for or against tends to be made by people who do not write the documentation, run the MDTs, or send the clinical correspondence the AI would actually change.
The clinicians who do those things have the clearest view of where AI would save real time and where it would not. If you are one of those clinicians the useful thing to do is make yourself visible in the local conversation. Not by lobbying for any specific outcome. By being the person in your service who has a clear view on what AI integration would actually change for clinical work.
That position is built one small action at a time. Trying the free tier. Asking the question about whether your service is in the initial rollout. Joining the meeting where the implementation gets discussed. Being the AHP who is not waiting to be told what AI is going to mean for their practice.
The 505,000 user rollout is a moment. What clinicians do with the moment is what determines whether the announcement was the beginning of something or the end of it.
In case you wanted to read: NHS Enghttps://www.england.nhs.uk/2026/06/500000-nhs-staff-to-get-new-artificial-intelligence-tools-to-help-free-up-more-time-for-patients/
That is all for Issue 10. Next week, the part of using AI that nobody teaches clinicians.
Clinically Intelligent drops every Wednesday. If you are not yet subscribed you can join free at clinicallyintelligent.com.
The AI tools discussed in Clinically Intelligent are consumer products. They have not been independently assessed by the author against DCB0129 or DCB0160 clinical risk management standards, and they may not be approved for clinical use by your employer. Before using any tool described in this newsletter in connection with your clinical practice, you must satisfy yourself that its use is permitted under your Trust information governance policy, your DSP Toolkit obligations, your professional registration requirements, and any applicable contractual terms with your employer. The author accepts no liability for use of any tool or workflow described in this publication. Patient identifiable information must not be entered into any consumer AI tool under any circumstances, irrespective of any guidance contained in this newsletter.