Clinic letters follow the same structure every time, and they still take an hour of someone's day. We build patient letter automation around the way your clinicians already dictate, so a transcript becomes a fully formatted letter, in your clinic's own template, ready for review. The clinical judgement stays exactly where it was.
No identifying patient data goes into anything we build. It formats, it doesn't decide.
Most private clinics have the same shape of problem. The clinical work finishes and the correspondence starts: new patient letters, GP follow-ups, discharge summaries, referral letters. Each one follows a template the clinic has used for years, each one takes real time, and each one has to be right.
Some clinics carry that on a medical typist. When that person retires or moves on, replacing them is hard and expensive, and specialist medical typing is a narrower talent pool than most owners expect. Other clinics have never had one, so the letters land on the clinician instead, and get written in evenings and gaps between appointments.
Either way it is the same bottleneck: a repeatable, highly structured writing job sitting on top of people whose time is worth more elsewhere.
The point of the design is that privacy is structural rather than procedural. The layer doing the drafting is never given a patient's name, date of birth or address, so there is nothing there to leak.
The clinician dictates the clinical content the way they always have, using whatever dictation software the clinic already runs. Patients are referred to without identifying details, so nothing identifying enters the process at this stage.
The transcript goes into a tool configured to your clinic's own templates: your headings, your section order, your house style, your standard paragraphs. It handles the different letter types you actually send and produces a complete draft rather than an outline.
Your admin team adds the patient's identifying details from your practice management system, and the clinician reviews and approves the letter exactly as they did before. The last word is still a human one.
The honest answer is that most of the risk in AI and clinical correspondence comes from feeding identifiable patient information into a system that had no business seeing it. So we design that possibility out. The drafting layer only ever sees anonymised clinical content dictated by the clinician. Identifying details are added afterwards, inside the clinic, by the people who already handle them.
That is a structural decision, not a policy someone has to remember on a busy Friday. It also means the tool sits alongside your existing systems rather than inside them, and it never becomes a place where patient records live.
If your clinic has its own information governance requirements, bring them to the call. We would rather scope around them properly than build something you then cannot use.
BZ Clinic in Surrey had a specialist medical typist retiring, and every letter followed a strict clinical template. We built a custom tool into their existing workflow: the clinician dictates as normal, and a fully formatted letter comes back in minutes instead of the hour it used to take. It was built and handed over in under a month, it is live today, and the clinic owns it outright.
That is why we are comfortable talking about patient letter automation as something we have done rather than something we offer. Read the full BZ Clinic case study →
Live today, and owned outright by the clinic.
"Marco is thorough, easy to work with, and always explains what he's doing and why."
Letter automation pays back where letters are high volume, highly structured and repetitive. It pays back much less where every piece of correspondence is genuinely bespoke, or where the clinic sends a handful of letters a month. We would rather say so on the call than build you something that saves an hour a quarter.
It also tends to matter more than clinics expect in specialties where demand already outstrips supply. If you do not need more patients, admin capacity is usually the only thing that actually changes your week.
Every clinic writes differently, so a build starts by learning how yours writes. There is no standard template we drop you into.
We look at the letter types you actually send and the templates behind them, using anonymised examples. Your existing correspondence sets the structural standard, not our idea of what a good letter looks like.
We build the tool around those templates: your headings, your section order, your standard wording, your tone.
Drafts are tested against anonymised letters and marked up by your clinicians. Their corrections feed straight back in, round after round, until the output is close enough to sign off with a read rather than a rewrite.
We train the people who will run it day to day, so it is not dependent on us. The clinic owns what we build.
Build time depends on how many letter types are in scope and how quickly review rounds come back. At BZ Clinic it was under a month, start to handover.
Letters are usually not the only thing eating the week. Enquiry routing, follow-ups and admin workflows sit in the same layer. See the full AI automation service →
The full build: the constraint, the three-layer workflow, and what changed. Read the case study →
Letter automation is scoped bespoke to each clinic, so it is quoted on its own rather than as a fixed package. See how we work →
No. The clinician dictates the clinical content themselves, exactly as they do now. The tool turns that dictation into a properly formatted letter in the clinic's own template. It does not diagnose, prescribe, advise, or add clinical content of its own, and every letter still goes out under the clinician's approval.
Identifying patient data never reaches the drafting layer. The dictation is anonymised, so names, dates of birth, addresses and GP details are simply never given to it. Those details are added afterwards, inside the clinic, from your own practice management system. The tool does not connect to your clinical records and does not store patient files.
That is the part most of the build time goes on. We work from your existing letters and templates, then run correction rounds where your clinicians mark up real drafts and those corrections feed back in. The goal is output you can approve with a read rather than a rewrite. It will not be perfect on day one, and the review step exists for exactly that reason.
It depends on how many letter types are in scope and how quickly review rounds come back. At BZ Clinic the build was done and handed over in under a month. The clinic owns what we build, and we train the people who will run it, so it does not depend on us afterwards.
No. The build works around the tools you already use rather than replacing them. Clinicians dictate the way they always have, and identifying details still come from your existing practice management system. If something in your setup makes that awkward, we would rather find out on the call than after a build has started.
One call. We will look at your letter volume and how they are produced now, and tell you honestly whether automating them is worth doing.
Book a discovery call →