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.
I have always had an eye for design. Not formally trained but I know my way around Canva and if something needs to look good I usually know how to make it look good. Which is exactly why certain things in the NHS frustrate me more than they probably frustrate most people.
I could not tell you what is on the posters on our walls. I have walked past them enough times that they stopped registering. And I know what is on them matters. Falls protocols. Hand hygiene. Pathways that exist for a reason. But they are walls of text with no visual pull. Nothing that makes my eye stop. So the information just does not absorb. I move past it. Most people do.
That is not a clinical problem. It is a design problem. And if you are a visual person like me you feel it every time you open an NHS document and see another wall of text where a clear diagram should be.
The Frailty Virtual Ward was up and running and we were looking at ways to improve different parts of the service. One of those was the patient leaflet. I designed one. The team liked it. The feedback was good. But getting it formally approved meant going through a review process that takes longer than most people expect — Trust branding rules, approved templates, sign off chains. The version I wanted to make and the version that could officially go out were going to end up looking very different. I understand why the process exists. It just means that by the time something reaches a patient it has often lost the thing that made it worth making.
What that experience showed me is that a lot of creative people in the NHS have quietly stopped trying. Not because the creativity left them. Because the effort of pushing it through stopped feeling worth it. You come in with ideas. The process says not quite. You try again. Eventually you start doing what everyone else does. You type it out. You print it. You hope someone reads it.
Most of the time they do not.
That is the gap Napkin AI tries to close.
This Week’s Tool
The carer guidance from Issue 04 works. Three tools, three jobs, one library of patient facing documents. But explaining it in writing meant readers had to work through several paragraphs before they could see what the system actually looked like.
I wanted a picture that would do the work the writing was struggling to do. Not a standard NHS workflow diagram either. Arrows in boxes. Readers have seen hundreds of those and stopped engaging with them a long time ago.
So I picked a metaphor first. Metamorphosis. A caterpillar at the start. A butterfly at the end. Three chrysalises between them. The metaphor maps to the workflow because clinical thinking develops the same way. You start with fragmented information. You pressure test it. You document it. By the time it comes out the other side it has been transformed into something a patient and their family can actually use. The chrysalis stage is the work nobody sees. The butterfly is what the family takes home.
I opened Napkin and pasted the text.

[Napkin AI suggestions panel showing the range of visual options available]
Napkin offered several visual options. Most were the standard workflow shapes I was trying to avoid. But a few were not. The metamorphosis was one of them. I picked it because it was the only option that actually did the work the metaphor was supposed to do.
That is the part worth focusing on. Not the final render. The expansion of what your thinking could look like.

[First Napkin output, the neon version with cyan and lime green palette]
That is what Napkin gave me first. The metaphor was there but everything else was wrong. Cyan. Lime green. Neon blue. The labels were louder than the illustration. It looked like someone had applied a generic style to my idea rather than expressed my idea through design.
When I saw that first version I knew the concept worked. The metamorphosis communicated exactly what I wanted it to. But I also knew that if the design did not earn someone’s attention the idea would sit there unread. A good idea in bad packaging is still ignored. I have spent enough time watching people walk past important information to know that.
So I pushed it.
I asked Napkin to apply pencil draft styling. Drop the neon palette. Use deep moss green and warm amber. Make the chrysalises faint sketches. Make the final butterfly the only fully coloured element on the page. The whole thing should read like a naturalist’s field notebook rather than a flowchart.

[Pencil draft version with sketched chrysalises and branch]
That is the pencil draft version. The branch reads as a sketched line. The chrysalises feel unfinished in the way that the middle of any real process feels unfinished. The butterfly is the only thing that is fully there.
Then I pushed further.

[Final naturalistic version, fully rendered metamorphosis with coloured butterfly]
That is the version I landed on. The metamorphosis fully rendered. The caterpillar, the chrysalises progressing through stages, the butterfly fully emerged at the end. The image now tells the reader what the workflow is before they have read a single word.
Fifteen minutes. A handful of prompts. That is what it took.
What Works and What Does Not
Three things Napkin does well.
The suggestions library is wide. When you paste text in you get a range of options. Most will be the standard shapes you have already seen. But there are usually one or two in there you would not have thought of. That is the part worth focusing on.
You can import your own branding. If your project or service has a visual identity you can pull elements of it in so the output starts to look like yours rather than Napkin’s defaults.
The free tier is generous. Five hundred credits covers most of what you would need to try it properly. You will not run out testing it.
Three things to know before you rely on it.
There is no phone editing. You need a desktop to work in Napkin seriously. The phone app is for viewing not building. If you work mostly on the move this matters.
You cannot edit text or colours after applying an effect. The pencil draft locks once applied. If you spot a typo afterwards you have to start again. Plan your text and palette before you apply anything.
The output is rarely finished. Napkin gets you to a strong draft fast. The final polish usually needs a second tool. Canva or wherever you do your design work will lift it from Napkin output to something closer to your visual language.
Information Governance
The use case in this issue is narrow. I used Napkin to turn my own newsletter content into visuals. No patient information. No Trust documents. That is the use case I have tested and the only one I am writing about.
Same rule as every issue. Nothing relating to a real patient goes into the tool. Not their name. Not their condition combined with their age and living situation. Not anything that would identify them to someone with reasonable knowledge of your service. If you want to create a visual that summarises a clinical workflow build it from a composite scenario as we covered in Issue 04.
A few things worth knowing about Napkin specifically.
Napkin is built on top of OpenAI and Google Gemini. When you upload content three companies are processing it not one. None of them are NHS approved data processors. Napkin is a US company on US infrastructure. Their privacy policy does not clearly state whether your content is used to train AI models. I could not find a clear opt out in the policy or in account settings. Until that is clearer the safe assumption is that anything you upload may inform AI systems somewhere in the chain.
Same reminder as always. Napkin is a consumer product. It has not been through NHS procurement or assessed against DCB0129 or DCB0160 — the clinical risk management standards NHS health IT systems are assessed against. Until your IG lead tells you otherwise keep real patient context and Trust confidential content out of it entirely.
If you are ever unsure ask your IG lead before you upload anything.
Where To Start This Week
Pick something you have already written that you wish had landed better. A pathway you explain the same way every time without it sticking. A teaching note your team forgot. An idea you have struggled to communicate to a colleague who learns differently.
Open Napkin. Paste the text. See what it offers you.
You do not need to ship the image. You just need to see your own thinking from outside it. Most of the time the act of looking at your own argument as a picture tells you something you did not know about the argument.
Ten minutes. One piece of writing. The work was already done. Napkin just shows you another side of it.
Prompts Worth Saving
One prompt this week. The one I used to take the metamorphosis from Napkin’s default rendering to something closer to my own visual language.
Apply pencil draft styling. Drop the neon palette entirely. Use deep moss green and warm amber as the only accent colours. Treat the middle stages as faint pencil sketches. Make the final element the only fully coloured thing on the page. The whole composition should read as a naturalist’s field notebook rather than a flowchart.
That prompt is doing one specific job. It is pulling the image away from Napkin’s defaults toward something with editorial intention. The structure is reusable. Replace pencil draft with whatever rendering style fits your work. Replace the colours with yours. Replace the field notebook framing with whatever metaphor your image is trying to land.
The prompt teaches Napkin to follow your taste rather than its own. The taste is still yours to develop.
Opinion
This issue is not really about Napkin. It is about something the tool made visible.
Most of us who work in healthcare are not visual communicators by training. We are trained in language. We write assessments. We write referrals. We hand over in words. When we are asked to communicate something important we reach for the format we know — text — and we produce something that looks like everything else. Another document. Another wall of text. Another thing that ends up in a folder nobody opens.
The irony is that visual communication is often more effective for the exact people we are trying to reach. Patients who find dense text hard to follow. Colleagues who retain pictures better than paragraphs. Families trying to remember what was said at a home visit after you have driven away. Visual formats reach those people in ways that written formats do not. We all know this. Most of us do not act on it because we assume it requires design training we were never given.
The people in the NHS who are naturally visual and creative are there. I know because I am one of them and I have worked with others. But the path from having a good design idea to getting it officially approved and distributed is long. Human feedback matters — the team response to the FVW leaflet was positive and that kind of early input is exactly the signal worth paying attention to before something goes through a formal process. The process exists for good reasons. It just means that by the time something reaches the person it was designed for it has sometimes lost the quality that made it worth making.
What AI tools like Napkin offer is a way to practise visual thinking without waiting for permission. You make something. You get feedback from the people around you. You iterate. Over time your eye sharpens and your ideas get clearer before they go anywhere near a formal process.
The creativity did not leave. It just needs an outlet.
In Case You Missed It
One story this week. It deserves the space.
Palantir and the NHS Federated Data Platform — The Financial Times and The Guardian reported that NHS England granted external staff including from Palantir unlimited admin access to identifiable patient data before it is pseudonymised.
An internal briefing acknowledged a “risk of loss of public confidence.” The change was approved anyway. The reason given was that requiring individual data access certificates was “too inconvenient” for external workers.
The Patients Association said patients had not been consulted. NHS England says access requires government security clearance and director level approval. Sources: Financial Times and The Guardian, May 2026.
Every time we sit with a patient and they share something sensitive they are trusting that what stays in that room stays protected by the system around them. That assumption is not just ethical. It is the foundation of the clinical relationship. How the NHS communicates what its data is used for and who has access to it will shape public trust for years to come. Most patients will not know this story exists. But it is their trust on the line.
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.

