Our approach to AI and probity

The short version

A scaffold, not a ghostwriter

The reflections in your portfolio must be your own — your experiences, your learning, your voice. ReflectMed is built around that principle. It prompts you with the right questions, then formats what youwrite into a Horus-ready structure. It does not fabricate clinical encounters, invent learning you didn't do, or write reflections on your behalf. Every draft is yours to review, edit, and stand behind before it goes anywhere near Horus.

What the guidance says

AI to structure is fine; AI to fabricate is not

The GMC's guidance on generative AI in portfolio and training applications is clear that AI may be used to help structure, organise and present reflective writing, as long as the content reflects the doctor's own knowledge and experience — a supplement to writing, not a substitute for it. The RCPCH's ePortfolio guidance draws the same line: using AI to structure and clarify your own account is acceptable, while fabricated encounters or mechanically cut-and-pasted, unedited output raise real probity concerns.

ReflectMed is designed to sit firmly on the acceptable side of that line — and to make it hard to cross by accident.

How ReflectMed is built to comply

Probity is a design decision, not a disclaimer

Scaffold, not ghostwrite. The AI prompts and formats; your own account leads. This is the core design of the product, not something bolted on afterwards.

A probity check before every save. Every reflection requires you to confirm a short set of attestations before it can be saved — that the account is your own, based on a real experience, and reviewed by you. If a supervisor ever asks, you can honestly say the entry is yours.

Grounded narratives only. When ReflectMed drafts a summary narrative, it draws exclusively from your own saved reflections and the steer you provide — it does not invent evidence or events to fill gaps.

Confidentiality guards.The other major risk the GMC highlights is patient-identifiable information finding its way into reflective writing. ReflectMed's anonymisation prompts and identifier checks are built to address it — our Privacy & data processing page sets out exactly how your data is handled.

Where we draw the line

What ReflectMed will never do

We will never generate assessor feedback, multi-source feedback (TAB/MSF), or self-ratings on your behalf — the guidance singles these out as unacceptable, and we don't build features there. We will never market ReflectMed as a tool that “writes your reflections for you”, because that framing is both untrue of how it works and exactly the wrong message. And we treat the confidentiality guards as something to keep permanently strict, not to relax for convenience.

Your part

The account stays yours

ReflectMed can structure and format, but it can't know what you actually experienced or learned — only you can. Read every draft, make it genuinely yours, and keep every entry anonymised in line with GMC guidance before you submit it. Using ReflectMed doesn't transfer that responsibility, and it doesn't guarantee any particular ARCP outcome.

One caveat

A plain-language summary, not legal advice

This page is a lay summary of how ReflectMed approaches AI and probity, and of published GMC and RCPCH guidance as we understand it — it isn't legal or regulatory advice, and the guidance itself is the authoritative source. If you're ever unsure about how AI use sits with your own portfolio, your educational supervisor and your medical defence organisation are the right people to ask. Questions for us? See our Terms and Privacy & data processing pages.