Eight Practical AI Uses for an Independent Optician

The most practical AI uses for an independent optician are recall messages, answering phone and web enquiries, explaining prescriptions and lens options in plain language, contact lens reorders, frame and product descriptions, review requests and replies, drafting referral letters, and analysing frame stock. Most need only a chat assistant and your existing practice software.

What decides which to try first is who does the work today. In a small practice the receptionist, the dispensing optician and the optometrist all lose time to different jobs, and AI helps each of them differently. Recalls and enquiries free the front desk; plain-English explanations help dispensing; referral drafting helps the optometrist. Pick the person with the worst backlog and start with their use, rather than trying all eight at once.

Follow me on Instagram@sagnikteaches

One rule that applies to all eight uses

Anything involving a patient's identity or clinical record happens inside software with proper data terms, such as your practice management system or a clinical tool with a signed agreement. General chat assistants are for templates, explanations and analysis with identifiers removed. In practice, a personal chat account should never see a patient's name, date of birth and clinical details together; a prescription on its own, with nothing tying it to a person, is fine for drafting an explanation.

Connect on LinkedInSagnik Bhattacharya

The running example is an illustrative independent practice with two optometrists, two dispensing opticians and one full-time receptionist, seeing around 35 eye tests a week and managing a few hundred contact lens wearers. The final sections cover clinical AI, which is a different kind of purchase, and a sensible order for the eight.

Subscribe on YouTube@codingliquids

1. Recall messages that patients actually answer

What AI does: writes recall messages that sound like your practice rather than a system, varies them by patient type, and, with the right tool, handles replies such as "can I come on a Saturday?"

Example (illustrative): three versions for three groups, written once and loaded into your practice software's recall templates.

• Routine eye test: "Hi [first name], it's about two years since your last eye test with us. Book a time that suits you here: [link], or reply and we'll find you a slot."
• Contact lens aftercare: "Hi [first name], your contact lens check is due this month. It keeps your prescription current and your eyes healthy, and we need it before we can supply more lenses. Book here: [link]."
• Children: "Hi, it's a year since [child's first name]'s last eye test. Children's eyes can change quickly, so a yearly check is a good idea. Book here: [link]."

How to start: ask a chat assistant for five versions of each recall message, pick the one that sounds most like your team, and paste it into your practice management system's template. Only then consider tools that handle replies automatically. Recall and booking automation in more depth is in how independent opticians use AI for recalls and bookings.

Watch-outs: recall intervals are clinical decisions. The message should follow the interval the optometrist set for that patient, not a default the AI assumed.

2. Answering phone and web enquiries

What AI does: answers the questions the receptionist hears all day: prices, availability, whether you do children's tests, contact lens trials, whether glasses are ready, how long a repair takes. It can be a chat widget on your website, an AI phone assistant, or simply AI-drafted replies to emails and messages that the receptionist sends.

Example exchange on a website chat (illustrative):

Visitor: "How much is an eye test and do I need to book?"
Assistant: "Our standard eye test is $65 and takes about 30 minutes, including a look at the back of the eye. We recommend booking, and you can pick a time here: [link]. If you're having problems with your eyes right now, please call us on [number] so we can see you sooner."

How to start: write a fact sheet: every price, opening hours, what each appointment includes, lead times for glasses and repairs, payment options, parking and access. The assistant answers only from that sheet.

Watch-outs: symptoms. A visitor who writes "I've suddenly got flashing lights and lots of new floaters" is describing something your optometrists will want to see urgently. The assistant must not reassure or wait for office hours; it should give the urgent instructions your optometrists have written, such as calling the practice immediately or, out of hours, the emergency route you specify. Have the optometrists write those rules, test them with made-up messages, and label the chat clearly as an automated assistant.

3. Explaining prescriptions and lens options in plain language

What AI does: turns the numbers and jargon of a prescription and lens recommendation into a short explanation a patient can read at home, which the dispensing optician checks and hands over.

Before, what many patients get: a slip reading "R -2.25 / -0.75 x 180, L -2.00 / -0.50 x 175. Rec: SV, 1.6 index, AR coat."

After, an illustrative AI draft checked by the optician: "Your prescription shows you are short-sighted, so distant things look blurred without glasses, and you have a small amount of astigmatism, which means your eyes focus slightly differently in different directions. Both are common and both are fully corrected by your glasses. We've recommended single-vision lenses, which are the same throughout, in a thinner lens material so they look neater in your frame, with an anti-reflection coating to cut glare from screens and headlights."

The prompt contains the prescription and the recommendation, never the patient's name. The detailed workflow, including explaining eye test results, is in explaining lens options and eye test results with AI.

Watch-outs: AI likes to promise. "These lenses will stop your headaches" or "you won't need stronger glasses" are claims nobody can make. The optician strikes anything that sounds like a guarantee.

4. Contact lens reorders and aftercare

What AI does: sends reorder reminders timed to when a wearer is likely to run out, handles simple replies ("yes, same again", "can I have six months this time?"), and flags anyone whose aftercare check is due before more lenses can be supplied.

Example flow (illustrative):

Trigger: 3 weeks before a wearer's supply runs out.
If aftercare is in date:
  Send: "Hi [first name], your contact lenses are due to run out around the
  20th. Reply YES to order the same again (3 months, $84), or tap
  here to change your order: [link]."
  On YES: create the order for the team to confirm and process.
If aftercare is overdue or due before the supply runs out:
  Send: "Hi [first name], your lenses are due to run out around the 20th, and
  your contact lens check is due too. Book your check here: [link]
  and we'll have your next supply ready."
  Do not create an order.

How to start: check whether your practice software or lens supplier's direct-delivery scheme already sends reorder reminders; many do. Add AI only for handling free-text replies that the standard system cannot parse.

Watch-outs: lens parameters. The AI should never change a lens type, power or replacement schedule, even if a wearer asks ("can I switch to monthlies?"). Those requests go to a person, because a change in lens usually needs a fitting. An illustrative holding reply the assistant can send: "Thanks, [first name]. Switching lens type usually needs a quick fitting check so we can make sure the new lenses suit your eyes. One of our team will call you tomorrow to arrange it, and we can send your usual lenses in the meantime if you're running low."

5. Frame and product descriptions

What AI does: writes website and social media descriptions for frames, sunglasses and lens options from the supplier's specification sheet, in your practice's voice, so your online range does not read like a catalogue copied from the manufacturer.

Supplier spec (as it often arrives): "Acetate full-rim, 52-18-145, havana, keyhole bridge, 5-barrel hinge, unisex."

AI draft (illustrative): "A classic rounded frame in warm tortoiseshell acetate, with a keyhole bridge that sits comfortably on most noses. Medium size (52 mm lens width), sturdy five-barrel hinges, and a shape that suits oval and square faces. Pop in and try it on, or book a styling appointment with our dispensing team."

How to start: paste ten spec lines at a time with your house style (length, tone, what to always mention) and ask for descriptions in a table you can upload. Add accurate image descriptions for the website at the same time, which helps visually impaired visitors. Virtual try-on and AI frame advice are a separate question, weighed up in whether AI frame advice and virtual try-on are worth it.

The same spec sheets feed social posts. An illustrative request, "three short posts introducing our new sunglasses range, friendly, no hype, mention prescription versions are available", returns drafts like: "New in: this season's sunglasses, from slim metal aviators to chunky acetate shapes. Every frame can be made up in your prescription, with polarised lenses for driving and days by the water. Come in and try them on." Check that every frame named is actually in stock and every lens option is one you supply.

Watch-outs: invented features. An AI that decides a frame is "lightweight titanium" when it is acetate creates a mis-description. Check every material and measurement against the spec sheet.

6. Review requests and replies

What AI does: sends a short review request after collection of glasses, and drafts replies to reviews for someone to approve.

Example request (illustrative): "Thanks for choosing us for your new glasses, [first name]. If you're happy with them, a quick review would really help a small independent practice: [link]. If anything isn't quite right with the fit, just pop in and we'll adjust them."

Example reply to a critical review (illustrative): "Thank you for taking the time to tell us. We're sorry your visit didn't feel right, and we'd like to put it right. Please call our practice manager on [number]." Note what it does not say: nothing about the reviewer being a patient, their prescription, their appointment or what was dispensed. Replying in public without confirming someone is a patient is the rule, and it goes in the assistant's instructions. There is more on handling harsh reviews in replying to negative reviews with AI.

How to start: send requests at collection time, by text, for a month, and count reviews before and after.

Watch-outs: posting replies automatically. Keep a person approving each one.

7. Drafting referral letters and clinical admin

What AI does: drafts referral letters, reports and letters to other clinicians from the optometrist's findings, in a consistent structure, for the optometrist to check and sign. It can also sit inside the test room as a clinical scribe, drafting the record from the conversation, which is how some optometry records systems now work.

Example (illustrative): the optometrist dictates, straight after the test: "Referral, routine. 68-year-old, gradual blurring in the left eye over six months. Visual acuity right 6/6, left 6/12. Pressures 16 and 17. Left: cataract, moderate, nuclear. Right: early lens changes. Fundus unremarkable both eyes. Patient keen for surgery; driving at night affected."

The AI draft sets that out as a letter with a clear reason for referral, the findings in a consistent order, the urgency, and the patient's wishes. The optometrist's checks: every number (acuities and pressures are where speech recognition slips), the laterality in every line, and the urgency, which must match what was said. A left cataract recorded as right in a referral is exactly the error that sends a patient down the wrong path.

How to start: use the letter or AI features in your practice software if it has them, or a clinical scribe with a signed data agreement. How scribes produce these drafts, and where their errors come from, is explained in what an AI scribe is and how it works.

Watch-outs: patient identifiers in general chat tools. Referral letters are clinical records; draft them inside systems built for clinical records, not a personal chat account.

8. Frame stock and buying analysis

What AI does: reads a sales and stock export and tells you what is selling, what is sitting, and where the money is tied up, in plain language.

Example (illustrative): the practice exports twelve months of frame sales and current stock by brand and price band (no patient data), and asks: "Which brands and price bands have the most stock relative to sales? Which frames have sat for more than nine months? Where should we cut or add on the next buying visit?" A typical answer:

Brand C: 64 frames in stock, 22 sold in 12 months. About 2.9 years
  of stock at current sales. 31 frames unsold for 9+ months.
Price band $250-$350: 38% of stock, 21% of sales. Overstocked.
Price band $120-$180: 18% of stock, 34% of sales. Often out of
  popular sizes (50-52 mm lens width).
Suggestion: reduce Brand C on next order, move part of the
  $250-$350 budget into $120-$180 in medium sizes.

How to start: export the data as a spreadsheet and try it with a chat assistant that reads files. Check two or three of its figures against the spreadsheet yourself before acting. For more on this kind of analysis, see whether AI can analyse a sales spreadsheet.

Watch-outs: AI arithmetic on large files can be wrong, and "stock" figures from practice software are sometimes out of date. Treat the analysis as a starting point for the conversation with the frame rep, not a buying order.

The fact sheet behind uses 2, 4 and 6

Three of the eight uses answer patients directly, and all three are only as good as the facts you give the assistant. A single fact sheet, kept in one place and loaded into a chat assistant Project, a chat widget or an AI phone tool, stops the assistant guessing. An illustrative extract:

FACT SHEET - last checked 1 September by [practice manager]
Eye test (adults): $65, 30 min, includes retinal photography.
Eye test (under 16s): $45, 30 min.
Contact lens fitting and trial: $80, includes first aftercare.
Contact lens aftercare: $40, needed every 12 months to supply lenses.
Glasses ready: usually 7-10 working days; varifocals 10-14.
Repairs: screws and nose pads while you wait; frame repairs sent
  away, 1-2 weeks.
Payment: card, cash, 0% finance on orders over $300.
Opening: Mon-Fri 9-5.30, Sat 9-1. Closed Sundays.
Access: step-free entrance, test room on ground floor.
URGENT SYMPTOMS: sudden flashes, new floaters, a shadow or curtain
  over vision, sudden loss of vision, painful red eye -> tell the
  person to call the practice now on [number]; out of hours, follow
  [the emergency route the optometrists specify].
Never: give clinical advice, predict prescription changes, or quote
  a price not on this sheet.

Three details make the sheet work. The date and owner at the top, because stale prices were the one failure the illustrative practice had. The urgent-symptoms line, written by the optometrists, not the receptionist or the vendor. And the "never" line, which gives the assistant a clear boundary when a patient asks something outside the sheet. The prices here are placeholders; use your own.

Update the sheet the same day anything changes, and ask the assistant a few test questions after each update ("How much is a children's test?", "I've got a curtain over my vision, can I book for next week?") to check it is reading the new version.

Clinical AI is a different purchase

Absent from the list: AI that interprets clinical images. Decision-support tools built into some OCT (optical coherence tomography) devices highlight possible abnormalities on retinal scans, and at least one, ZEISS's CIRRUS PathFinder, has been studied with optometrists. These can be valuable, but they are clinical equipment decisions, not admin tools. They come with their own regulatory status, evidence and training needs, and the optometrist remains responsible for the interpretation. Evaluate them with your professional body's guidance and the manufacturer's clinical evidence, not alongside a chat assistant subscription.

A month at the illustrative practice

The practice started with three of the eight, chosen because they relieved the receptionist, who was the bottleneck:

1. Recall templates (use 1): rewritten in an afternoon and loaded into the practice software. Responses to the first recall wave rose, which the practice tracked as bookings per hundred messages.
2. Enquiry replies (use 2): the receptionist pasted email and message enquiries into a chat assistant with the fact sheet saved in a Project, then checked and sent. About 40 minutes a day became about 15.
3. Prescription explanations (use 3): the dispensing opticians generated plain-English explanations for every new spectacle wearer and every lens upgrade. Patients took the sheet home, and "what did she say about the coating?" phone calls dropped.

Cost for the month: one individual chat assistant plan at about $20. The practice deliberately did not buy an AI phone assistant until it knew how many calls were routine; the receptionist kept a tally sheet for two weeks first. Referral drafting (use 7) was next on the list, but only once the practice software supplier confirmed how its AI features handled patient data.

One thing went wrong. The fact sheet listed last year's price for a children's test, and the assistant quoted it in three email replies before anyone noticed. The fix was a named owner for the fact sheet and a date at the top showing when it was last checked.

How to tell whether each use is working

Each use has one number worth tracking, and most of them can be read from your practice software or a simple tally.

• Recalls: bookings per hundred recall messages, compared with the old template over the same season. A rise of even a few bookings per hundred is worth having across a whole recall list.
• Enquiries: time from enquiry to reply, and the share of enquiries answered the same day. Tally a fortnight before and after.
• Prescription explanations: "what did they say about my lenses?" calls after collection. Ask the receptionist to mark them on a sheet.
• Contact lens reorders: wearers who lapse, meaning no order and no aftercare within three months of running out. This is where lost contact lens revenue hides.
• Frame descriptions: visits to the frames pages on your website and "saw this online" mentions at the desk.
• Reviews: new reviews per month and your average rating.
• Referral drafting: optometrist time per letter and corrections needed per letter.
• Stock analysis: the share of frames unsold for more than nine months, checked at each buying visit.

Pick the measure before switching the use on, record a baseline, and check again after a month. If the number has not moved, the use is either set up badly or not worth your time, and both are worth knowing before you add the next one. It also keeps the conversation honest if a supplier later claims its AI product will transform your recalls: you will already know your own figures.

An order for the eight

The pattern is simple: start with the uses that need no patient data and no new software, prove the time saving, then move to the ones that touch records or need a new tool, once you have the data terms in writing. Most independent practices can get uses 1, 3, 5 and 6 working within a month for the price of one chat assistant plan, and that month tells you whether the rest are worth the extra effort.

Questions independent opticians ask about AI

Which of the eight should a two-optometrist practice try first?

Usually recalls and enquiry handling, because they touch the most patients and the results show within weeks as bookings. Plain-English prescription explanations come next, because they are cheap and improve every dispense. Leave stock analysis and referral drafting until the team is comfortable checking AI output, since both need a careful eye.

Is it safe to put patient records into ChatGPT to write letters?

Not on a personal plan, and not with names or identifiers on any general assistant unless your data terms and rules allow it. For referral letters, use the AI features in your practice software or a clinical scribe with a data agreement. For templates and explanations, general assistants are fine with no patient details included.

Do I need special AI software, or will a chat assistant do?

Most of the eight work with a general chat assistant plus your existing practice software. The exceptions are phone answering, which needs an AI receptionist or phone service, and anything that reads the patient record, which should happen inside software with the right data terms. Start with what you have before buying anything optical-specific.

Further reads

• How to Write Call Scripts and Escalation Rules for an AI Receptionist — Write the call script and escalation rules for AI phone answering.
• Patient Reviews and AI Replies: Staying Within Confidentiality — Reply to patient reviews without breaching confidentiality.
• How to Check Your Margins Product by Product With AI — Check margins on frames and lenses with AI.
• AI Alt Text: How to Generate and Check Image Descriptions — Write accessible image descriptions for frame photos.
• What to Ask Before Buying Any AI Tool for a Medical Practice — Questions to ask before buying any clinical-adjacent AI tool.
• 10 Admin Tasks a Small Clinic Can Hand to AI This Month — More admin a small clinic can hand to AI this month.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: vendor information on AI-assisted OCT interpretation, including a published reader study of ZEISS CIRRUS PathFinder with optometrists; OpenAI and Anthropic pricing pages.