Typing is the slowest way to produce veterinary notes, and it isn't close: controlled research puts speech input at three times typing speed with fewer errors. Dictation fixes the input speed but still leaves you composing every sentence yourself, after the consult, from memory. Ambient AI scribes remove the composition step too — you consult normally and edit a draft — and the best clinical evidence available (from human medicine; veterinary trials don't exist yet) links them to about 0.9 fewer hours of after-hours documentation per day and a burnout drop from 52% to 39% within 30 days. The practical answer: type only to edit, dictate when you want word-level control in specific PIMS fields, and let an ambient scribe carry the full consult-to-SOAP workload. Here's the evidence behind that, and the honest caveats.
The three ways a vet note gets written
Every clinical note is produced by some mix of three jobs: remembering what happened in the consult, composing it into clinical language, and entering it into the record. The three methods differ in which jobs they take off your plate:
- Typing leaves all three with you. You remember, you compose, you type.
- Dictation (speech-to-text) automates entry only. You still remember and compose — you just speak the sentences instead of typing them.
- Ambient AI scribes automate all three for the first draft. The consult itself is the input: the tool listens, transcribes, and structures a draft note. Your job shifts to reviewing and correcting.
That framing explains almost every result in the research below — and why moving from dictation to an ambient scribe is a bigger jump than moving from typing to dictation.
Typing: the baseline everyone underestimates
Typing feels fast because it's familiar. It isn't fast. In a controlled study by Stanford University, the University of Washington and Baidu, speech input was 3.0x faster than typing in English, with a 20.4% lower error rate. And that study measured pure text entry — it didn't count the part where you sit down at a keyboard hours after the consult and try to reconstruct which ear it was.
That reconstruction cost is the hidden tax of typed notes in practice: the composing happens at the end of the day, when the details have faded and the exam room context is gone. The result is a queue — and we've written about where that queue ends up: notes that follow you home.
Where typing genuinely wins: editing. Correcting a word, restructuring a sentence, tweaking a template — for short, precise changes, keyboard beats voice. Which is exactly the role typing keeps in a scribe-based workflow.
Dictation: three times faster input, same composition burden
Veterinary dictation has been around for decades — first with human transcriptionists, now with veterinary-vocabulary speech-to-text that types wherever your cursor is, including straight into PIMS fields. The Stanford numbers are the strongest argument for it: same note, roughly a third of the entry time.
What dictation doesn't change is the composing. You still formulate every sentence of the S, O, A and P yourself — and you still do it after the consult, from memory. That's why dictation tends to shine in specific spots rather than as a whole-workflow answer:
- Fragments into specific fields, where you want exact control ("grade 2/6 left systolic murmur")
- Referral letters and callbacks, where the wording is the work
- Clinicians who think out loud fluently and dislike editing drafts
It's telling that vendors who started in veterinary dictation — Talkatoo being the clearest example — now sell an ambient scribe alongside it, positioning dictation as one input mode within a broader documentation assistant rather than the destination.
Ambient AI scribes: what the studies actually show
There are no published randomized trials of AI scribes in veterinary clinics yet — every number a vet-scribe vendor quotes (including ours) is a vendor claim, not independent evidence. The closest independent data comes from human medicine, where the tools and the problem (a clinician talking to a patient, a structured note due afterwards) are directly analogous. Two studies matter:
The JAMA productivity study (1,800 clinicians using AI scribes vs 6,770 controls across five US hospitals, over two years) found modest average gains: 13 fewer minutes of EHR time and 16 fewer minutes of documentation time per day, plus about half an extra patient visit per week. The crucial detail is the dose effect: clinicians who used the scribe for more than half their visits saw twice the EHR-time reduction and three times the documentation-time reduction. Averages were dragged down by people who barely used the tool — adopt it properly or it does little.
The JAMA Network Open burnout study (263 ambulatory clinicians, six US health systems) measured what changes in the first 30 days of ambient scribe use: burnout fell from 51.9% to 38.8%, clinicians reported 0.9 fewer hours per day of after-hours documentation, and cognitive load during note-writing dropped alongside improvements in attention on the patient.
Read together, the two studies say something more specific than "AI scribes save time": the clock-time savings are real but moderate, while the biggest measured effects are on when the work happens (after-hours time) and how it feels (burnout, cognitive load). For a profession where work-life balance is the strongest predictor of wellbeing, that's arguably the point.
The honest caveats
- Review time is real time. An ambient draft still needs your read-through and corrections — the studies above measure net gains, but a sloppy review habit turns time saved into risk transferred. The draft is documentation assistance; the approved note is yours, and so is the clinical judgment in it.
- The evidence is borrowed. Human-medicine results should transfer reasonably to a structurally identical workflow, but "should" is doing work in that sentence. Until veterinary trials exist, treat all quantitative claims — from any vendor — as hypotheses to test in your own clinic.
- Accuracy varies with your room. Multi-speaker consults, barking, accents, and non-English consult languages all stress transcription differently. The dose-effect finding cuts both ways: the tool only pays off if it works well enough that you actually use it on most consults.
- A trial answers all of this in two weeks. Every major veterinary scribe offers a free trial. Record five real consults, count your edits, and check the note lands in your PIMS without manual copy-paste — we've written a full comparison of the current options with published pricing and trial terms.
Which method fits your workflow
- Stay with typing if your caseload is light, your notes are genuinely short, and your records are already done before you leave. (If that's you, you're rare — and this article wasn't for you.)
- Choose dictation if you want word-level control, you write lots of referral letters, or you want speech input inside specific PIMS fields without changing anything else about your workflow.
- Choose an ambient scribe if your bottleneck is the SOAP notes themselves — especially if they're currently written hours after the consult. This is the only method that attacks the composition and memory cost, not just typing speed.
- Combine them — which is what most scribe-era workflows settle into: ambient capture for consults, dictation for letters and odd fields, keyboard for edits.
One more variable that's easy to miss: language. If you consult in Danish, German, Norwegian or Swedish, test any tool in that language specifically — transcription quality outside a product's core languages varies enormously. (It's the reason our own product, Journal.vet, is built natively multilingual for European practices.)
Frequently asked questions
Is dictation dead now that ambient scribes exist? No — it's been demoted from "the answer" to "a mode." Word-level control still matters for letters, corrections, and precise field entry. What's fading is dictating entire SOAP notes sentence by sentence when a scribe can draft them from the consult itself.
Do I talk differently in the exam room with an ambient scribe? Mostly no — the tools are built for natural consult conversation. Many vets do adopt one useful habit: verbalizing key findings during the physical exam ("temperature 38.5, mild tartar on the left carnassial") so they land in the transcript. Clients generally read this as thoroughness.
How much review does an AI draft actually need? Plan for a full read of every draft, heavier edits in the assessment than anywhere else, and decreasing correction time as you tune templates. If after a two-week trial the edits still outweigh the savings, that tool (or that tool in your room) isn't ready — try another before concluding the category isn't.
Will an AI scribe make my notes better, or just faster? Faster is the measured claim; better is up to you. A scribe reliably captures what was said — but the structure and reasoning standards of a good note don't change with the tool. If the fundamentals are shaky, start with what belongs in each SOAP section.