Everyone knows AI is here to stay. The fear mongering has not changed that and the evidence is clear. For healthcare professionals working inside a system that is genuinely stretched — burnt out staff, outdated technology, processes that eat hours that should be spent with patients — AI is not a threat. It is one of the biggest opportunities we have to improve the system.
The issue is the system not the people. Every clinician I know is curious. Most are quietly trying things already. The frustration is with the pace. The NHS is cautious by nature and in many ways that caution exists for good reason. But caution has a cost. While the rest of the world is figuring out how to use these tools practically and safely, many of us are still manually updating spreadsheets, writing the same referral forms we were writing five years ago, and waiting for someone above us to tell us what is allowed.
That conversation is not coming anytime soon. And honestly — you do not need it. Because there are things you can do right now, today, this week, that are safe, practical and will genuinely give you time back. Real tools. Real workflows. Things you can open on your laptop or phone after reading this and actually use before your next shift.
Clinically Intelligent is for the healthcare professional who spent two hours this week manually updating a spreadsheet with patient stats that AI could have compiled in seconds. For the doctor who clerked five patients before lunch and still had discharge summaries waiting. For the nurse who stayed late chasing results that a better system would have flagged automatically. If you want to use AI to get the mundane work done faster so you can spend more time actually doing the job you trained for — this is for you.
Welcome to Issue 01.
Information Governance
Before we get into any tools, let me address the elephant in the room. Am I actually allowed to use AI at work?
The honest answer is yes. With one rule that overrides everything else.
Never put information into an AI tool that could identify a patient. No names. No NHS numbers. No dates of birth. No addresses. Nothing that could connect the content back to a real person. That is it. That is the line. Stay on the right side of that line and you are safe.
That is my personal rule and it is the one I will apply to every tool and workflow in this newsletter. No identifiable patient information. Ever. Full stop.
I will be honest with you. I am not an information governance expert and this newsletter is not legal advice. The rules vary between Trusts and they are still trying to catch up with how fast AI is moving. If you are ever unsure about something specific at your organisation the best person to ask is your IG lead.
What I can tell you is how AI works in practice and how you can use it practically — as a clinician who is figuring this out the same way you are. Practical tools and real workflows you can actually use.
This Week's Tool
Claude — For Your Own Professional Work
Claude is the tool I use the most and the one that has changed my workflow the most.
Think about something as simple as taking a social history. Whether you are in a patient's home or on the ward, your brain is already running through your last few visits and you are relying on a piece of paper and whatever you can hold in your head. Misplace that paper and you are trying to reconstruct a conversation from memory. We have all been there.
Now this is my workflow. I open a blank Word document during the assessment and type as the patient talks. I can stay completely focused on what they are saying without losing my train of thought the moment my pen starts running out of ink. I capture everything. And honestly — we should all be moving this way. The NHS Long Term Plan commits to going paperless and a Word document on your laptop is a small but meaningful step in that direction.
Once I am back at the office I check that nothing identifiable has made it into the document. Then I upload it to Claude, which already has my CGA or social history template loaded along with a clear prompt telling it exactly what to do with the information — more on prompts later.
It fills the template. I review it. I paste it into the EPR.
What used to take significant time and mental energy and honestly, sometimes suffered in quality because you had six others to get through — now takes minutes and the detail is there every time. No gaps. No missing information because you were tired or rushing.
Prompt of the Week — CGA and Social History in Minutes
This is the prompt I use most. If you read the Claude section earlier you already know the workflow. This is the engine behind it.
There are two versions below. One for CGA which is used widely across frailty, acute, and community settings. One for social history which is more broadly applicable across specialities. Use whichever fits your practice and adapt the template to match your service.
Before you start — tell Claude who you are and where you work. Something as simple as: "I am an occupational therapist working in an acute NHS Trust in the UK." This does two things. It improves the accuracy of the output because Claude understands your clinical context. And it ensures the language is correct — without it you might get mobilize instead of mobilise, or terminology that does not match UK clinical practice. The more context you give Claude at the start of a conversation the better everything that follows will be.
Copy the template of your choice into Claude along with the instruction at the top. Then paste in your typed assessment notes underneath and let Claude do the rest.
Version One — CGA
Download the full CGA template here — https://docs.google.com/document/d/1KsQikUEp2nPaeDh5_Fz6jvVsLoOGXAGeFmgNsc5fFkw/edit?usp=drivesdk
The instruction to paste into Claude:
You are a clinical documentation assistant helping a healthcare professional complete a Comprehensive Geriatric Assessment. I am going to paste my typed assessment notes below. Please use the information in my notes to fill in the following template as accurately as possible. Only use information that is present in my notes. If a section has no information available write unknown. Do not invent or assume any clinical details.
Then paste your typed assessment notes underneath and let Claude fill the template.
Version Two — Social History
Download the full Social History template here — https://docs.google.com/document/d/1bHcwOjmIWSjhZZoRpCQOv7lFp8EVev8Q_1rG5dAOa8E/edit?usp=drivesdk
The instruction to paste into Claude:
You are a clinical documentation assistant helping a healthcare professional complete a social history. I am going to paste my typed assessment notes below. Please use the information in my notes to fill in the following template as accurately as possible. Only use information that is present in my notes. If a section has no information available write unknown. Do not invent or assume any clinical details.
Then paste your typed assessment notes underneath and let Claude fill the template.
A few things worth knowing. If the output is not quite right do not start again from scratch. Just tell Claude what to fix. You can keep refining in the same conversation until it is right. Once you have a version that works for your service save it somewhere accessible. And one conversation per patient — start fresh for each assessment and keep it clean.
Two things to remember for both versions. Remove any identifiable patient information from your notes before you paste them in. And always review the output before it goes into the EPR. AI is your first draft not your final signature.
Opinion
The NHS is one of the largest organisations in the world. It employs over one million people and serves millions of patients every year.
Which is exactly why the pace of AI adoption is so frustrating.
The NHS regularly cites burnout and unsustainable costs as existential threats. And they are right. Burnt out staff leave. Burnt out staff make mistakes. Burnt out staff spend their careers doing work they trained for whilst drowning in documentation, stats, MDTs, and administrative tasks that follow them home. There is a direct link between that administrative burden and the burnout crisis. Anyone who has worked a full clinical caseload knows it.
AI will not fix everything. But it can fix some of it. The mundane, repetitive, time consuming tasks that eat into clinical time every single day — documentation, summarising, chasing, logging, reporting — these are exactly the kinds of tasks AI handles well. Time saved on those tasks is time given back to patients. It is also time given back to the clinician. And that matters.
Different industries have already figured this out. Finance, law, education, retail — AI is embedded in how they operate and the people doing the work are better for it. The NHS is not a different species. The same tools that save a lawyer two hours of document review can save a Band 7 two hours of clinical notes.
The good news is that this does not require a policy change or a procurement committee or an eighteen month pilot scheme. It requires individual clinicians deciding that their time is worth protecting and taking the first step themselves.
That is what this newsletter is for. Not to wait for the NHS to catch up. To make sure you do not have to.
That is all for Issue 01. Every week I will bring you something practical you can use and a view on where this space is heading. Next week we go deeper on a tool that can transform the way your team accesses information. If someone you know would find this useful, pass it on.