The NHS Clinical AI Fellowship: Everything You Actually Want to Know Before Applying
So you've heard about the NHS Clinical AI Fellowship and now you're down a rabbit hole at midnight, trying to figure out if it's actually worth applying. Sound familiar?
Whether you stumbled across it in a newsletter, heard a colleague mention it, or have been eyeing it for months, there's a good chance you still have more questions than answers. The official information only gets you so far, and what you really want is the unfiltered, practical breakdown that tells you what this fellowship actually involves day to day.
That's exactly what this post is here for. We're cutting through the vague programme descriptions and giving you the stuff that matters: who the NHS Clinical AI Fellowship is genuinely suited for, what the application process looks like in practice, what fellows actually spend their time doing, and whether it could realistically fit into your career right now.
By the end, you'll have a much clearer picture of whether this is the right move for you, and if it is, you'll know exactly how to put your best foot forward. Let's get into it.
What the Fellowship Actually Is and Why It Exists Now
Let's be honest with ourselves for a moment. How many of us have sat in a procurement meeting, been handed a glossy brochure about some new AI diagnostic tool, and quietly thought: "I have absolutely no idea how to evaluate whether this is safe or just impressive-looking"? If that resonates, you are not alone, and more importantly, you are not the problem. The gap is structural.
The NHS Fellowship in Clinical AI was founded in 2021 by the London AI Centre, with backing from Health Education England, and is hosted by Guy's and St Thomas' NHS Foundation Trust. It now operates under a unified national brand at nhsfellowship.ai, having grown from a London-centric initiative into something with genuine UK-wide reach. Scotland is already on board, with NHS Education for Scotland funding candidates and two Scottish fellows completing the programme in August 2025.
What makes this fellowship significant is the explicit problem it was designed to solve. It is described as "the first systematic route in the UK to acquire the relevant skills in clinical AI deployment", which is a fairly striking statement when you sit with it. The programme was created directly in response to the Topol Review's call for dedicated clinical AI training, and it sits squarely within the UK government's ambition to become a global AI superpower, including the NHS's own stated goal of becoming the most AI-enabled workforce in the world by 2035.
This is not a pilot. The fellowship is currently in its fifth cohort (August 2026 to August 2027), with Cohort 6 applications opening in October 2026. That trajectory, from inaugural cohort to a mature nationally branded programme in five years, tells you something meaningful about both the demand and the commitment behind it. If you have been waiting to see whether this programme would stick around before investing your time in applying, the answer is fairly clear now.
The Two Days a Week Commitment: What That Really Means
The commitment is two days per week for twelve months, running alongside your existing clinical work or training rather than replacing it. That sounds manageable on paper. In practice, it deserves some serious thought before you hit apply.
If you are a registrar, two protected days per week represents roughly 40 percent of a standard working week. That is not trivial. The programme requires an Approval in Principle from someone with genuine decision-making power over your training pattern, and you will not even be invited to interview without it. This exists for good reason: protecting those days in a busy rota, especially across on-calls, study leave, and ARCP requirements, needs real buy-in from your training programme director and deanery. It is worth having that conversation early and honestly, not as a formality.
If you are a consultant, the maths may look friendlier. Depending on your job plan, you may already have SPAs or management sessions that can be restructured. Two days carved from an existing mix of clinical and non-clinical PAs is a different proposition to a registrar pulling two days from a 1-in-6 rota.
The good news is that the work itself is grounded in reality. Fellows are matched to existing NHS clinical AI projects and embedded within multidisciplinary teams, so you are not studying AI in the abstract. You are applying it to live clinical workflows, which makes the learning stick.
The honest framing is this: it is demanding but designed to be doable. Think carefully about your rota, your training stage, and your personal circumstances before committing.
Money: How Salary Cover Actually Works
Let's talk about money, because pretending it doesn't matter would be doing you a disservice.
Salary cover is provided for nationally and regionally funded posts, and that single fact removes what is otherwise the biggest practical barrier stopping most registrars from even considering the programme. The mechanism is straightforward: your employing trust or deanery is reimbursed at 0.4 FTE unbanded salary for the time you spend on the fellowship. You can check the specifics in the Cohort 4 Role Description, which sets out the financial architecture clearly. Your department does not quietly absorb the cost. Your rota coordinator does not have to perform miracles. The funding is structured so that releasing you is genuinely feasible for most clinical teams.
The distinction between nationally funded and regionally funded posts matters more than the programme literature sometimes lets on. Nationally funded posts are awarded centrally. Regionally funded posts depend entirely on whether your deanery has committed money, and that commitment is not uniform across the NHS. Some regions are well set up; others are not. International applicants, and anyone in a post without dedicated funding, will need to explore bespoke arrangements via the fellowship directly.
Here is the practical takeaway: do not wait until Cohort 6 applications open in October 2026 to ask these questions. Deanery conversations move slowly. Contact your educational supervisor and Training Programme Director now, ask explicitly whether funded posts exist in your region, and if the answer is unclear, email gstt.aifellowship@nhs.net directly. A programme backed by the Topol Review and cited in the NHS Long Term Workforce Plan does not run on goodwill alone; the financial architecture is real and worth pursuing.
The Curriculum: Four Things They Want You to Understand by the End
The FCAI curriculum is built around four formal learning domains, and they fit together more logically than you might expect.
1. AI Fundamentals
This is not a computer science degree compressed into a fellowship year. The goal is to give you enough conceptual and technical grounding to engage critically with AI tools in clinical practice. That means understanding how models are trained, what training data actually represents, and where bias creeps in before a tool ever reaches a patient. Fellows can work through self-paced data literacy resources and introductions to Python as used in health analytics, but nobody is expecting you to become a data engineer. What the programme does expect is that you stop nodding along in vendor meetings and start asking better questions.
2. Clinical AI Regulations and Standards
Here is where things get genuinely practical. AI software in the UK is treated as a medical device under MHRA frameworks, and the implications of that for procurement, deployment, and post-market surveillance are significant. The curriculum covers information governance, NHS commissioning, and UKCA marking considerations. One Scottish graduate specifically highlighted AI quality assurance and post-market surveillance as crucial learning, which tells you something about where real-world risk actually sits.
3. Validation and Evaluation
Can you critically appraise a clinical AI study the way you would a drug trial? This domain teaches you how. Fellows learn to design evaluation frameworks, assess performance in their specific patient population, and identify when published accuracy metrics simply do not transfer to local clinical reality. The University of Glasgow's Digital Health Validation Lab is a named delivery partner here, which reflects how seriously the programme takes rigorous, independent assessment.
4. Integration and Systems Impact
This is arguably the most underappreciated domain. Deployment failure is almost never about the algorithm. One histopathology trainee fellow described progress being slowed by existing clinical frameworks and stakeholder scepticism, which is exactly why this strand exists. Fellows are embedded in live NHS clinical AI teams, not reading case studies about them. The curriculum treats post-deployment monitoring as a continuous obligation, not a box-ticking exercise.
Delivery happens through bespoke masterclass workshops at NHS centres of AI excellence, alongside your embedded project work. Partners include the NHS Digital Academy and the University of Glasgow's Digital Health Validation Lab, bringing real institutional weight to what could otherwise feel like on-the-job learning without structure.
What a Fellowship Day Might Actually Look Like
Once you're in the programme, the work is genuinely applied from day one. Fellows are matched to existing NHS clinical AI projects and embedded within a multidisciplinary team, which means you're not sitting in lectures or working through hypothetical case studies. You're in a real clinical environment, contributing to something that has already been scoped and is actively moving forward.
A typical fellowship day might look something like this: reviewing outputs from an AI-assisted diagnostic tool alongside clinical colleagues, flagging where the model's confidence scores don't align with clinical judgement, then heading into a project meeting with data scientists and clinical informaticians to discuss evaluation methodology. Later in the day, you might be drafting a section of a quality improvement report on an AI-supported pathway, or preparing a presentation for a departmental governance meeting. It's varied, contextual, and occasionally humbling in the best way.
What helps enormously is the supervision structure. You're working under a dedicated clinical AI supervisor throughout, which gives you a senior clinical lens on what can otherwise feel like a very technical environment. NES-supported graduates have described the supervisory relationship as central to their development, particularly when navigating governance and post-market surveillance questions.
The portfolio value here is real. Fellows are supported to publish and present their work, and the programme has a dedicated publications section precisely because fellowship outputs are expected to be substantive. Formal QI projects, clinical evaluations, and supported publication pathways all feed directly into your CV and your post-fellowship credibility.
The practical immersion model is deliberate, not incidental. The programme wants you to understand what deployment and evaluation actually feel like in a live NHS context, not in a classroom, and that distinction matters enormously once you're back in your substantive role.
Who the Fellowship Suits and Who Should Probably Wait
Let's be clear about something upfront: this is not a radiology fellowship that has been rebranded. Historically, fellows have come from radiology, neurology, paediatrics, and general practice, but the programme is genuinely open to any clinical specialty. If you work in emergency medicine, oncology, respiratory, or any other discipline where AI tools are landing in your workflow, you belong in this conversation as much as anyone else.
The candidate who tends to thrive here is someone who has been in enough AI-adjacent meetings, procurement discussions, or journal clubs to have developed a healthy sense of what they do not yet know. Not someone who wants to learn Python. Not someone who thinks they can skip the clinical bit. The sweet spot is the digitally curious clinician who is ready to stop nodding along and start contributing meaningfully to how these tools get deployed safely.
In terms of career stage, the programme fits registrars at ST3 and above particularly well. You need enough clinical grounding to contextualise what an algorithm is actually doing in a real patient pathway. Consultants looking to formalise skills and step into clinical AI leadership roles are also well placed, though the eligibility specifics around training numbers mean it is worth checking the current criteria carefully rather than assuming identical terms apply.
What it probably does not suit is anyone hoping for a lighter clinical period. As covered earlier, this adds two days of structured commitment to your existing job. It rewards those who are already stretched but organised, not those running on empty.
For medical students following along, formal eligibility is not there yet. But understanding this fellowship's structure now, while the applicant webinar for Cohort 5 is still available to watch, is genuinely useful career planning. Clinical AI literacy is becoming baseline, not specialist.
The Training Grade Question: ARCP, Deanery Sign-Off, and Approval in Principle
If you are a registrar in a training programme, this section is the one you need to read twice. The fellowship is structured to run alongside training, and the mechanism that makes that possible is called Approval in Principle. This is a formal submission made by your Training Programme Director (TPD), not by you directly, confirming that your deanery or Health Education England regional team supports your participation. It is listed as a hard eligibility criterion in the role description, not a courtesy step. That means before you even think about writing your application, you need to have a real conversation with your TPD and get them on board.
What Approval in Principle does not do is automatically protect your ARCP outcome. The fellowship is designed to sit at 0.4 FTE alongside your remaining clinical work, but your training programme's clinical exposure targets, competency sign-offs, and progression milestones do not pause while you spend two days a week on AI projects. Your educational supervisor needs to understand exactly how the fellowship days will interact with your rota and your portfolio, and that conversation should happen before you submit an application, not after you have been offered a place.
Some specialties are more accommodating than others. The East of England region, for example, requires applicants to be at least ST3 before applying, reflecting a judgment that more junior trainees may not have the bandwidth to absorb LTFT activity safely. Other regions and programmes will have their own informal thresholds. The Cohort 4 role description is explicit that workforce and geographical eligibility decisions are made at regional level and are not uniform across the NHS.
If you are based in Scotland, your route runs through NHS Education for Scotland rather than an NHS England regional team. NES collaborates directly with GSTT to fund Scottish candidates, and two Scottish fellows completed the programme in August 2025 under this arrangement, demonstrating that the pathway is real and functional.
The core practical message is straightforward. Designed to run alongside training is not the same as guaranteed to run smoothly alongside training. The administrative groundwork sits with you, and it needs to start well before the application deadline closes.
What Happens After: The Clinician-Technologist Identity
Completing the fellowship doesn't mark the end of something. It marks entry into a professional community that still feels genuinely close-knit, because it is. The NHS Fellowship in Clinical AI LinkedIn network reflects a growing but still intimate group of clinicians who share something relatively rare: formal, structured experience in AI deployment rather than just a passing familiarity with the technology. That distinction matters more than it sounds. There is a meaningful difference between a clinician who attended a conference about AI and one who spent a year embedded in a live NHS deployment project, supervised within a multidisciplinary team, working through regulation, validation, and integration in real clinical workflows.
The career directions emerging from this cohort are varied and still evolving. Alumni are moving into clinical AI lead roles within trusts, contributing to AI procurement and governance committees, building academic collaborations, and in some cases informing national policy conversations. These are not roles with established job ladders. They are being shaped by the people filling them, which is both the challenge and the appeal.
This reflects something broader happening across NHS medicine. The clinician-technologist is becoming a recognised career archetype, not a niche add-on. For those think
ing about professional identity beyond a single specialty, the fellowship is one of the very few NHS-sanctioned routes to build a cross-cutting expertise that travels across organisations, systems, and borders. The AI community landscape shows parallel academic routes emerging, but none currently offer the same combination of live deployment experience, structured curriculum, publication support, and policy grounding in a single programme.
What this means practically is that alumni are equipped to interrogate AI procurement decisions from a clinically credible position. As trusts face increasing pressure to evaluate and commission AI tools with limited in-house expertise to scrutinise them properly, that skillset is becoming genuinely valuable, and increasingly sought after.
Application Timeline: Cohort 5 Is Live, Cohort 6 Opens October 2026
Cohort 5 is currently running, covering August 2026 to August 2027, and if you missed that window, the application deadline passed on 15 December 2025. No need to dwell on that. The more useful thing to know is that Cohort 6 applications open in October 2026 for an August 2027 start, which means right now sits in an ideal preparation window. Use this time well: have an honest conversation with your educational supervisor, establish whether your deanery holds funded posts, and work through the nhsfellowship.ai apply page in detail rather than skimming it the night before applications open.
One data point worth sitting with: the Cohort 5 applicant webinar in November 2025 attracted 1,711 views on YouTube, from a channel with only 83 subscribers. That is not passive background noise. That is over a thousand clinicians actively researching the application process, which tells you something real about the level of organic competition you are entering.
Funded posts are finite, and the programme does not publish exact numbers per cohort, so building your case before October is a sensible investment. Relevant QI work, involvement in digital health projects, or even demonstrable engagement with clinical AI literature all strengthen an application meaningfully.
For colleagues outside the UK, the programme now accepts international applicants through a dual fellowship pathway based at Guy's and St Thomas', though bespoke funding arrangements apply rather than the national salary cover available to NHS applicants. The fact that this route exists at all reflects how seriously the UK's model is being watched globally.
Why This Matters for NHS Medicine Right Now
The NHS is not slowly drifting towards AI. It is already deep into it. Radiology departments are running AI-assisted reporting tools. Emergency triage workflows are being shaped by algorithmic decision support. Administrative systems are using automation to manage referrals and documentation. This is not a future scenario; it is the current operational reality for most trusts. The problem is that much of this procurement and deployment has happened faster than the clinical workforce has been prepared to engage with it critically.
That is not an accusation. It is a structural gap, and it has been formally recognised for years. The 2019 Topol Review named dedicated clinical AI training as a strategic workforce priority, and the fellowship is the most substantial national-scale answer to that recommendation currently running. It does not just tick a policy box either; it sits within the NHS Long Term Workforce Plan and is framed by the government's ambition to make Britain a genuine AI superpower by 2035.
The sharper question for any clinician reading this is not whether AI is coming but whether you are an informed participant in how it gets used or simply a downstream recipient of someone else's decisions. Those are meaningfully different professional positions.
There is also something bigger at stake here around professional identity. Being a great doctor in 2026 increasingly means understanding the evidence frameworks and systems shaping your practice, not just your specialty. The clinician-technologist is becoming a recognisable career archetype, and programmes like this are the formal pathway into it.
At Clerked, we think a lot about what modern medical identity looks like. The doctor who can evaluate an AI tool with the same rigour they bring to a clinical trial is part of that story, and it is one worth paying attention to.
Key Takeaways Before You Apply
Here is what you genuinely need to carry forward from everything covered above.
This is the UK's most structured route into clinical AI leadership. Two days a week, twelve months, running alongside your existing post. That structure is the point. It is designed to produce clinicians who can lead, not just observe.
Salary cover exists, but you need to ask the right people. Nationally and regionally funded posts carry cover, but availability varies by region. Contact your deanery directly and ask whether a funded place exists for Cohort 6 before you assume the finances will be a barrier.
Registrars: start the Approval in Principle conversation now. The October 2026 application window will arrive faster than you expect. Your educational supervisor needs time to engage with this properly.
The curriculum makes you a credible clinical voice, not a data scientist. AI Fundamentals, Regulations and Standards, Validation and Evaluation, Systems Impact. Four domains, one clear purpose.
Start building relevant experience regardless of which cycle you target. A QI project with a digital angle, reading clinical AI literature regularly, and one honest conversation with a current fellow will do more for your readiness than any last-minute preparation.
Conclusion
The NHS Clinical AI Fellowship is not for everyone, but for the right person, it is a genuinely career-defining opportunity. Here is what to take away from everything covered above.
First, the fellowship suits clinicians who want to shape how AI is used in healthcare, not just observe it. Second, the application rewards clarity and genuine motivation over technical credentials alone. Third, the day-to-day experience is hands-on, collaborative, and far more practical than a traditional academic programme.
If this sounds like the path you have been looking for, do not wait for the perfect moment to apply. Start drafting your application now, reach out to current or former fellows, and back yourself.
The clinicians who will define the future of NHS AI are applying today. Make sure you are one of them.