AI Scribe for Optometrists: Exam Notes, Templates and a Test Plan

An optometrist can use an optometry-specific scribe such as Doctora, which lists $199 per doctor a month billed annually, the scribe built into their own EHR where one exists, or a general scribe such as Heidi (free plan) for the narrative. Test any of them on 20 to 30 real exams over two weeks.

The question that decides most of the others is how the note reaches your chart. A tool that writes fields per eye into your EHR saves a different amount of time from one that hands you a block of text to paste. So before comparing prices, find out which of the three routes below your EHR supports: a cheap scribe that leaves you retyping refractions and pressures can cost more time than it saves.

Follow me on Instagram@sagnikteaches

Paste or write-back: the difference that sets your time saving

Eye-exam notes are mostly structured: acuities, refraction, pressures, slit lamp and fundus findings per eye, diagnoses and a plan. A scribe that only produces prose leaves you copying numbers into boxes one at a time. A scribe that writes into the fields does the filing for you. Neither is universally better, and the right one depends on your EHR.

Connect on LinkedInSagnik Bhattacharya

Say a practice has two doctors and a tech who pretests. If the scribe pastes text, each exam might save two minutes. If it writes into the fields, it could save more, because per-eye entry is where the typing goes. But a write-back tool that is in beta for your EHR may need more checking, so the saving on paper is not always the saving on your screen.

Subscribe on YouTube@sagnikteaches

Options as the vendors describe them

I read these pages in November 2026. Prices are list prices and change, so confirm before you commit.

Two things in that table deserve a second look. First, "live" and "beta" are the vendor's words, and beta means the integration is built but still maturing; Doctora says it gives at least 30 days' notice before changing beta pricing or availability. Second, the price comparison depends on what you already pay for. If your EHR's scribe is included in your plan, the question becomes whether a separate tool is worth paying for on top.

An exam template you can adapt, filled in

Whether you build it in a scribe or paste the structure into instructions, a clear template matters more than the brand of tool. Below is a structure for a comprehensive exam, followed by an illustrative filled-in draft with the kind of values a scribe should either hear or leave as "not stated".

COMPREHENSIVE EXAM TEMPLATE
Reason for visit / chief complaint: patient's words
History: ocular, medical, medications, allergies, family history
Entrance vision: unaided and aided, distance and near, OD OS OU
Refraction: sphere, cylinder, axis, add, OD and OS separately
Best-corrected acuity: OD OS OU
Pupils, motility, visual fields: findings or "not tested"
Pressures: method, value OD OS, time
Anterior segment (slit lamp): lids, conjunctiva, cornea, AC, iris, lens
Posterior segment: disc, cup-to-disc ratio, macula, vessels, periphery
Diagnoses: each with laterality
Plan: glasses, contact lenses, drops, imaging, referral, recall
Education given: what was explained, questions asked

Illustrative draft (fictional patient):
Reason for visit: "Distance is blurry when I drive at night."
History: Wears glasses for 15 years. No eye pain. Dry eyes in
  the evening. Father has glaucoma. No known allergies.
Refraction: OD -2.25 -0.75 x 175, add +1.75
            OS -2.00 -1.00 x 005, add +1.75
Best-corrected acuity: OD 20/20, OS 20/20
Pressures: OD [not stated], OS [not stated]
Anterior segment: Mild lid margin debris OU. Cornea clear OU.
Posterior segment: Cup-to-disc ratio 0.5 OD, 0.5 OS. Macula flat.
Diagnoses: Myopia with astigmatism OU; presbyopia; dry eye symptoms.
Plan: Updated glasses; discussed night-driving lenses; lid hygiene;
  recall in 12 months; glaucoma family history noted for monitoring.

Notice that the pressures line reads "not stated". A scribe should not guess a value it did not hear, and a template that forces a blank is safer than one that lets the model fill in something plausible. If your tech enters pressures during pretest, the line should come from there, not the scribe. A good write-back tool keeps what the tech entered and shows who entered what, which Doctora describes for its integrations.

A prompt for a general scribe, with sample output

If you try a general scribe with custom instructions, tell it exactly how to treat measurements. This prompt is a starting point, tested on ordinary language and meant to be adjusted after your first week.

"Draft a comprehensive eye exam note using the headings below.
Rules: Use only what was said in the visit. Write every
measurement per eye as OD and OS. If a measurement was not
spoken, write [not stated]. Never estimate. Do not add diagnoses
that I did not say. Keep patient quotes short. List any advice
I gave under 'Education given'. At the end, list any item you
were unsure about under 'Please check'."

Illustrative output extract:
Please check:
- Axis OS heard as "five" or "fifteen"; recorded as 005.
- Patient mentioned eye drops but the name was unclear.

The "Please check" list is the most useful part of that prompt. It turns the tool's uncertainty into a to-do list instead of a confident wrong answer. You will find that real drafts flag things like unclear drug names, ambiguous axis values and notes that depended on a gesture rather than words.

What goes wrong in eye-exam drafts

• OD and OS swapped. A patient says "my right eye is worse" and the draft puts the finding on the wrong side. Check every per-eye line.
• Axis and sign errors. "Minus one-twenty-five" heard as "plus", or "axis ninety" as "nineteen". Compare refraction values with the phoropter or autorefractor display.
• Cup-to-disc ratios and pressures invented. If you did not state them, they should be blank. A plausible 0.3 with a normal pressure is the most dangerous kind of invention.
• Findings from your usual routine written in. Drafts can "normalise" an exam: "no abnormalities" for a structure you did not examine.
• Drug names and doses. Eye drops share similar names. Confirm each drug, strength and frequency yourself.
• The tech's and doctor's voices mixed. If both are recorded, check who said what.

Each of these is a thing to score in the test below, so that you know the rate rather than guessing.

Consent wording for the exam room

The scribe listens to the exam, and the patient has to know. Ask before the visit if you can, for instance on the intake form, and again briefly in the room. Keep the wording short, because patients are often already anxious about eye tests.

On the intake form:
"We use an AI tool that listens during your eye exam and drafts
the doctor's notes. The doctor reviews every note. Your recording
is deleted [when the note is confirmed / after __ days]. You can
say no, and your exam will be exactly the same. Please tick:
[ ] I agree   [ ] I would rather not"

At the start of the exam:
"I use an AI note-taker during exams, so I can look at you and
not the screen. Is that all right? You can say no at any point."

A few practicalities specific to optometry. A technician's pretest conversation is part of the visit, so decide whether the scribe listens then and tell the patient. Children and parents both speak, so ask the parent, and speak to the child about it too. If you serve patients in states where recording needs everyone's consent, the map and form in the consent form tutorial will help. Keep a no-scribe route: type or dictate afterwards, and log the choice in the chart.

Three practices, three different answers

The "best" tool changes with the practice. Here are three illustrative set-ups, with the reasoning rather than a verdict.

A solo optometrist on an EHR that has its own scribe. She starts with the built-in scribe, because her EHR vendor's page says it is included with the exam module and it already writes to her chart. Her test compares it with Doctora's free first ten charts, on the same ten exams. If the built-in scribe misses refraction detail, she has a reason to pay. If it does not, she saves the subscription and the extra login.

A two-doctor practice on an EHR that Doctora lists as beta. They would be early users of the write-back, so they ask the vendor for a call, a named contact and the 30-day change notice in writing. They test on ten exams before paying and ask two questions of their own: does a technician's pretest survive untouched, and what happens when the doctor edits a value in the draft after the write-back? A beta integration can be excellent and still need more checking in the first month.

A practice on an EHR with no write-back anywhere. They use paste. A general scribe such as Heidi drafts the history, counselling and plan, and a technician enters per-eye numbers as usual. The time saving is smaller but the risk is lower, because the scribe never touches measured values. They score the tool on the narrative sections only.

Contact-lens fits and medical visits need their own templates

A comprehensive exam template stretches badly over other visit types. A contact-lens follow-up has different content from a red-eye visit, and a template that tries to cover both tends to produce padded notes. Keep short separate templates and test each one.

Contact lens follow-up (illustrative filled line set):
Lens: [brand/type], OD [base curve/diameter/power], OS [...]
Wear time: 10 hours a day, 6 days a week; replacement as advised.
Comfort: "fine until about 8 pm, then dry".
Vision: OD 20/20, OS 20/20 with lenses.
Fit: Centred, adequate movement OU. [not stated for edge lift]
Care: Solution changed to [name]; hand hygiene reviewed.
Plan: Re-check in 3 months; remove lenses and call if red or painful.

Medical visit (red eye) headings:
Onset and symptoms | Discharge, pain, light sensitivity |
Vision | Findings per eye | Diagnosis (with side) | Treatment
and drop schedule | Warning signs given | Follow-up

Medical visits deserve extra scrutiny, since a missed warning sign matters more than a missed lens brand. In the medical template, add a rule to the instructions: "Always record any red-flag symptoms the patient mentioned, in the patient's own words, and list which warning signs I gave." Then check the draft for those two items first.

Who does what when a technician pretests

Many optometry practices split the exam: the tech takes history and pretest, and the doctor does the refraction and health check. A scribe fits into that flow in one of three ways, and the choice affects accuracy and consent.

1. Doctor only. The scribe starts when the doctor enters the room. The tech's entries stay as they are in the chart. Simplest to consent and to review.
2. Tech and doctor, one recording. The scribe hears both, which adds the tech's voice to the consent and makes it harder to tell who said what.
3. Tech first, then doctor, two recordings. Two sessions per patient. Useful when the tech takes a long history, but double the consent checks and the review.

Most practices in a first test choose option one. It limits the scope of what could go wrong and keeps measured values where staff enter them. Once that works, you can widen it. Doctora's page describes keeping what the tech entered and showing who entered what; if your tool does something similar, check it on your own staff first.

Getting the room set up for the microphone

Many early disappointments with a scribe come from audio, not from the model. The exam room is full of competing noise: the phoropter clicks, a slit-lamp joystick, a fan, a corridor. Place the device between you and the patient at about desk height, not on the instrument stand. Test it with a ten-minute mock exam and read the draft for words the tool got wrong, since misheard numbers tend to cluster near the noisiest steps. If your patients are often in a darkened room during the fundus check, remember that the conversation continues while you work, so remind them to speak up. Change one thing at a time, and re-test.

A two-week test, day by day

Run the test on real exams but keep your own notes going until you trust the tool. Two weeks is enough because the first days expose set-up problems and the later days show steady-state quality.

The review checklist and scoring sheet

Use the same checklist for every draft, then add up the results. A filled-in scoring row looks like this for one illustrative exam.

REVIEW CHECKLIST (per draft)
[ ] Every OD/OS value matches what I measured or was told
[ ] Refraction signs and axes match the instrument printout
[ ] Pressures and C/D ratios are mine, or marked [not stated]
[ ] Nothing "normal" is recorded for a structure I did not examine
[ ] Drops: name, strength, frequency checked
[ ] Diagnoses carry laterality and match my findings
[ ] Plan, recall and referral match what I told the patient
[ ] Codes suggested: I chose, not the tool

SCORING ROW
Exam 7 | Comprehensive | Review minutes: 3 | Own note minutes: 8
Wrong values: 1 (axis OS) | Invented: 0 | Missing: 1 (drop name)
Signed with under two edits? No

After 20 to 30 rows, three numbers matter most: the proportion of drafts you would sign with under two edits, the average review time against your own writing time, and the count of invented findings. One invented finding in thirty is a template problem to fix. Several is a reason to stop and ask the vendor.

Reading the results without fooling yourself

Two traps catch practices at the end of a trial. The first is the halo of the good day: three smooth exams in a row feel like proof, so score all of them, not the ones you remember. The second is the bad-day verdict, where one irritating error in a rush colours everything. Let the sheet decide.

Say a practice scores 24 exams and finds 15 signed with under two edits, six needing three to five edits, and three where the draft was worse than starting fresh. That is 62.5 per cent clean, which is a reasonable base, and the three poor drafts are all medical visits. The sensible reading is not "the tool fails" but "use it for routine exams and keep typing medical visits until the medical template improves". A split decision like that is common, and it is a perfectly good outcome for a first fortnight. Write the decision down with the date and the numbers behind it, so that in six months, when the vendor has updated the product, you can run the same sheet and compare.

What it costs, in a practice with two doctors

An illustrative sum: two doctors each see 14 exams a day for 20 days a month, so 280 exams per doctor. At Doctora Core's listed $199 per doctor a month, that is $398 for the practice, about $0.71 per exam. If the tool saves three minutes per exam, each doctor gains 14 hours a month, so the cost works out near $14 for each hour saved. If the saving is one minute, the cost per hour saved triples. Time your own exams before you trust any of these numbers.

Heidi's free plan costs nothing and its pricing FAQ says scribing never uses up the plan's usage allowance, though advanced templates sit on the paid plan at $110 per user a month billed annually. If your EHR's built-in scribe is included in your subscription, the extra cost is zero, and you only pay in the time it takes to review drafts. Compare all three on the same exams, not on brochures.

What I could not verify for you

Two points are worth asking about directly. Whether your specific EHR supports write-back from a given scribe is a fact only the vendor can confirm for your version and setup. And whether your EHR vendor's own scribe covers refraction and slit-lamp fields in the way you need is something to test, not assume. If you want a broader view of how a decision rule starts from your EHR and region, see the medical practice decision rule, and for the write-back question in a neighbouring profession, the dentist write-back tutorial. For six kinds of note error that carry over to eye exams, the note errors checklist is a good companion.

If you would like a hand planning which two tools to test and how to score them, that is the sort of thing my AI implementation consultation covers on a 1:1 call.

Optometry scribe questions

Does an AI scribe work with my optometry EHR?

It depends on the tool and the EHR. Some write fields straight into the chart and others only give you text to paste. Ask each vendor for your exact EHR by name, and ask whether that integration is live or still in beta. Do not assume a logo on a list means it is ready.

Can the scribe record the technician's pretest as well?

Only if you set it up that way, and that adds another voice to the consent question. Many practices let technicians enter pretest data as normal and have the scribe cover only the doctor's exam, so that measured values stay where staff entered them.

Should it suggest billing codes?

Suggestions can save time, but the doctor chooses the final code. A scribe does not know what you did unless it was said or entered, so check every suggested diagnosis and exam code against what the visit actually contained.

Is a free scribe good enough for exam notes?

It might handle the history and conversation well, but optometry notes are full of per-eye measurements and structured fields. A free general tool can draft the narrative while you type the numbers. Test whether that split is faster than your current method.

How many exams should a test cover?

Twenty to thirty across two weeks is enough to see patterns, provided you include comprehensive exams, contact-lens fits, medical visits and a few awkward cases such as children or patients who need an interpreter.

Further reads

• Twofold AI Medical Scribe for Clinicians: Trial Checklist and Cost — A trial checklist for another scribe you might put on the shortlist.
• AI Medical Scribe for Clinicians: Shared Templates in a Group Practice — How to share templates when several doctors use the same tool.
• What Is an AI Scribe and How Does It Work in a Consultation? — Plain-English explainer of how a scribe works in a consultation.
• AI Scribe Security: Audio, Retention and BAA Vendor Questions — Vendor questions on audio, retention and agreements.
• AI Scribe for Psychologists: Notes, Reports and Consent Checks — Consent checks when the visit content is especially sensitive.
• AI Scribe Costs Compared: What Small Clinics Pay in 2026 — Price context for small clinics weighing scribes.
• AI for Optometrists: Contact Lens Aftercare and Reorder Messages — Draft contact lens aftercare and reorder message sequences with AI, with stop rules and a route for discomfort replies to staff.
• Optometry Insurance Verification: A Front-Desk Workflow With AI — A front-desk routine for checking vision and medical eye insurance before the visit, with a call script, note template and safe uses of AI.
• Optometry Referral Letter Template: Draft With AI, Then Check — A referral letter template for GPs and ophthalmologists, a safe AI drafting method, and a pre-send checklist for laterality, numbers and urgency.
• AI in an Optometry Practice: A Week-by-Week First-Month Plan — A four-week plan for an optometry practice: phones, recalls, exam notes and marketing, with what to measure and what to leave out.
• GDPR in an Optical Practice: Which AI Tools May Touch Records — A scored 12-point checklist for vetting any AI tool before it touches eye-test records, with four tools worked through and the paperwork behind each.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Doctora home and pricing pages; RevolutionEHR's AI Scribe product page; Heidi pricing and compliance pages. Prices as read on 4 November 2026.