Orthodontic AI Receptionist: Consults and Broken-Bracket Calls

Yes, with limits. An AI receptionist can book new-patient orthodontic consultations from a fixed script and pass broken-bracket calls to staff with the caller's details. It should not triage, advise or judge urgency. Test every booking in your software, and route anything about pain, injury or a poking wire to a person.

The split that works is simple. Booking a consult is a closed task with a short list of facts to collect. A broken bracket is an open clinical question that sounds minor on the phone and sometimes is not. So script the first completely and script the second as a handoff: collect, tell the caller what happens next, notify a named person, log it. Then prove both on test calls before you switch the phones over, because a script that reads well on paper can still book a child under a parent's name.

Follow me on Instagram@sagnikteaches

Why orthodontic calls differ from general dental calls

A general dental line is dominated by recalls, hygiene visits and the occasional toothache. An orthodontic line has a different mix: parents of children asking about a first consult, adult patients asking about aligners, braces patients rescheduling adjustment visits, and the calls nobody enjoys, which are "my bracket has come off" and "a wire is sticking into my cheek". Each type needs a different answer, and the AI should know which are its jobs.

Connect on LinkedInSagnik Bhattacharya

The rule behind the table: the AI may do anything that is a booking or a message, and nothing that is judgement. See fee, emergency and booking rules for a dental chatbot for the wording to set before launch.

Subscribe on YouTube@sagnikteaches

The consult-booking script, written out

A script has a better chance of working than an open prompt because it limits what the tool can say. Write yours in steps and give the tool the exact words for the parts that matter. Below is a filled-in example (illustrative) for a practice that offers a complimentary first consultation of 30 minutes.

OPENING
"Hello, this is the automated assistant for [practice name]. I can book a
consultation or take a message for the team. How can I help?"

IF the caller wants a first consultation:
1. "Is the consultation for you or for your child?"
2. Ask for: patient first name, age, parent or guardian name (if under 18),
   best phone number, email.
3. "Has a dentist referred you?" (store yes/no and referrer name; do not
   judge referral quality)
4. "Which of our locations suits you: [Site A] or [Site B]?"
5. Offer the next three consult slots at that site. Never offer a slot the
   system does not show.
6. Confirm: "That's [day] at [time] at [site], about 30 minutes. A parent or
   guardian needs to come with anyone under 18. You'll get a text confirming it."
7. Ask: "Is there anything the team should know before you arrive?" Record
   the answer verbatim. Do not comment on it.

NEVER: quote treatment prices, say whether braces or aligners are suitable,
predict treatment length, comment on photos or X-rays, promise a particular
orthodontist.

What the output looks like when it works (illustrative): a booked 30-minute consult at the right site, the parent's phone number in the right field, "referred by general dentist, name given" noted, and a confirmation text. What to fix on first test calls: a tool that offers "next Tuesday" without saying the site, a tool that books a child's consult under the parent's name, and a tool that answers "is it too late for my 15-year-old?" with encouragement. The script's NEVER line is there for the last one.

If your consult is free, a follow-up after the visit is where conversions are won; that is in following up free consultations. Payment options belong in that written follow-up from the team, never in the AI's phone script.

The broken-bracket handoff, step by step

The goal is that a caller with a broken bracket never hears advice from a machine and never waits without knowing what happens next. The handoff has five steps, and the AI does only the first three.

1. Recognise. Trigger on words and phrases: bracket, wire, band, loose, broken, poking, sharp, cut, sore, swollen, bleeding, hit, knocked. Do not make the AI decide which words matter; give it a list and tell it that any match goes to the handoff.
2. Collect. Name, date of birth or parent's name, phone number, site, and one open question: "Can you tell me in a sentence what has happened?" Record the reply word for word.
3. Tell. "Thank you. I am passing this to the team now. Someone will call you back by [time rule]. If you have severe pain, swelling or an injury to your face or mouth, please call [emergency number set by the practice] or seek urgent care." The last line is fixed wording that the practice's clinician approves, not generated by the AI.
4. Notify. Send a text and email to the named person on duty with the caller's words, number and the time. If the practice has a task list in its software, create a task there.
5. Close. A staff member calls, gives advice, offers a repair visit. The AI never closes the loop on its own.

The "time rule" is the part owners underestimate. Pick one: during open hours, callback within 30 minutes; outside open hours, callback by 09:00 next morning, and a transfer to the on-call number if the caller says pain or injury. Write it into the script, tell the tool, and measure it.

An illustrative handoff message that arrives on the duty phone:

URGENT-CHECK  [Site B]  14:12
Caller: [first name] [surname], parent of [child first name], age 13
Phone: [number]
Said: "One of the brackets came off at lunch and the wire is catching on his cheek."
Callback promised: by 14:42
AI gave no advice. Logged in call notes.

A staff member reading that can ring in a minute and decide what to say. The AI-written version should never contain the words "just", "probably" or "it is likely", because those are judgements. Keep the message quoted and factual.

Multi-site rules that stop callers booking the wrong chair

Every extra site gives the tool another set of hours, slot types and phone numbers to mix up. Write a short rule sheet and give it to the vendor.

• Announce the site every time. The tool states the site name and address before confirming a slot.
• One site per booking. Never offer slots across sites in one list unless the caller asks.
• Each site has its own hours, its own consult length, its own escalation number and its own after-hours rule.
• Named providers are optional. If a caller asks for a specific orthodontist, the tool books only if that provider is shown at that site, otherwise it takes a message.
• Test each site separately, with its own set of five calls, on different days.

For a group wondering whether to run all calls through one centre, when a group practice should centralise calls sets out the trade-offs.

Named tools to test, and what to ask each

Search results for this topic are dominated by receptionist vendor pages. Treat each as a claim to test. One example with a published page is Orthia, built for orthodontic practices. Its home page says it answers calls, books appointments and verifies insurance during the call, and names Cloud 9, Dolphin Management (V10 and above), Ortho2 Edge, OrthoTrac (V14.3 and above), Dentrix Ascend and Greyfinch. Its integrations page adds two sentences worth reading twice: read and write capabilities vary by practice-management system and version, and for systems without write-back it creates scheduling tasks for your team to confirm. As of November 2026 its pricing page lists Standard from $399 a month per clinic (after-hours and overflow calls) and Premium from $499 a month per clinic (full 24-hour cover), month to month with no per-call fees, and custom pricing for larger groups. Standard includes "emergency triage with staff tasking"; neither page mentions broken brackets by name, so ask the vendor how it handles one, and whether "triage" means it judges urgency (which your script should not allow) or only routes by keyword.

Weave, a dental phone and messaging platform, also sells an AI Receptionist; its product page says direct booking works "for qualified Dental PMS only" without naming which, so ask whether your orthodontic software qualifies. For general dental phone tools that also serve orthodontists, Weave, Zo or a generic agent compares the field. For each tool, ask the five questions below and write down the answers.

1. Which of my practice-management versions do you write to, and which only read?
2. Can I define a list of words that always trigger a handoff?
3. Can I fix the callback wording so the tool cannot improvise it?
4. Do I get a transcript or recording of each call, and for how long is it kept?
5. Can I leave month to month, and what happens to my number?

A worked example: the numbers for a two-site practice

Illustration only. A two-site orthodontic practice gets 90 calls a day across both sites. Staff miss about 18 of them at lunch and at close because both front desks are busy with patients in the chair. Say 4 of those 18 are new-consult enquiries and 2 are broken-bracket calls. If the AI receptionist books three of the four consults and the practice's average consult converts at 60 percent, the expected number of new starts from the previously missed consults is 1.8 a day from the calls alone, before any other change. At 21 working days that is about 38 starts a month. Do not trust this: your missed-call numbers come from your phone log, your conversion rate from your own records, and the tool will not book all of them. The sum shows which two inputs to measure for a month before you buy: missed consult calls, and consult-to-start rate. Put the price beside it: at Orthia's listed starting prices, two clinics would pay from $798 a month on Standard or $998 on Premium, before any custom quote for a group.

The other cost is the two broken-bracket calls a day. Before the AI, they waited for a free hand. After the AI, they get a message to staff in under a minute. If staff call back within the promised 30 minutes on 90 percent of them, the practice has a service level to report to itself. Measure that number weekly from the message timestamps and the call log. The method for tracking returns over the first quarter is in measuring an AI receptionist's return.

Test calls to run before the phones go live

1. A parent asks for a consult for a 12-year-old at Site B on a Saturday. The site does not open on Saturdays; the tool should say so and offer weekdays.
2. An adult asks whether aligners would work for them. The tool should say a clinician will discuss this at the consultation and offer to book, with no opinion.
3. A caller says "my bracket fell off and my gum is bleeding". The handoff should trigger, notify the person on duty, and give the fixed urgent-care line.
4. A caller with a heavy accent or poor connection says a name the tool cannot spell. It should read the name back and ask for the spelling, and never book with a guess.
5. A patient asks to cancel their adjustment and rebook "after the holidays". The tool should offer the next slot after the date and warn staff if the interval exceeds your limit.
6. A caller asks "is this a real person?" The tool should say it is an automated assistant and offer a message to the team.

Fix the script after each failed call, then rerun all six. Keep a log with date, call, result and fix. If the vendor cannot change the behaviour, that is your answer on the contract. When a booking goes wrong after launch, what to do when an AI receptionist gets a booking wrong covers the recovery message.

What the callback sounds like once staff pick up

The handoff is only half the job; the callback decides how the family feels about the practice. Give the person who rings back a short guide, written by the clinical lead, so the call is quick and consistent. A typical shape, illustrative:

1. Name the caller and the issue back to them: "I have your message about [child first name]'s bracket."
2. Ask the two questions your clinician wants answered, such as whether there is pain or bleeding and whether the wire is the problem.
3. Give the advice your clinician has approved for that situation, or book the earliest repair slot.
4. Record the outcome in the patient's notes and close the task.

Keep the approved advice on a single page that staff can read to callers; the AI never sees it. Review it every year with the clinical lead, because treatment habits and materials change.

Adults, teenagers and parents: different callers, same script

Orthodontic callers include parents booking for children, teenagers calling themselves and adults exploring aligners or braces. The consult script should ask who the consult is for in the first question and branch from there. Adults may want evening slots and a quiet way to ask about cost; teenagers often call with a parent in the background; parents ask who will be in the room. The AI should treat all three the same way: collect the same fields, give the same confirmation, and pass cost and suitability questions to the team.

One field is easy to forget: the guardian's name for under-18s, which many practices need before treatment starts. Put it in the script as a required step and test it. Another is consent for text messages. Ask for it once, record it, and do not send reminders to a number that has not agreed.

A short before-and-after for a Friday afternoon

Before: it is 16:40 on a Friday. Both front-desk staff are with patients and a parent calls to say a bracket has come loose before a weekend trip. The call rings out, goes to voicemail and is picked up on Monday. After: the AI answers, collects the details, says a team member will call by 17:10 and sends the message to the duty phone. A staff member rings at 16:55, gives the approved advice and books a repair slot on Monday morning. The parent never reached a clinician in the first version and reached one in the second. The only thing the AI did was take down the facts and move them quickly, which is exactly the job.

The same Friday shows why the tool must know your closing time. If it keeps promising a 30-minute callback at 17:45, the promise is empty. Give it the hours of each site and a different rule for the last hour of the day.

Recording, disclosure and the data the tool keeps

Voice tools usually record calls, transcribe them and keep the text for a period. Children's details are involved, so read the vendor's data terms before you sign. Ask: where are recordings stored, how long, who can see them, will the vendor sign a business associate agreement or processing agreement if your rules require it, and can you delete a recording on request. Put a short disclosure line at the start of the call. The exact wording and legal duties depend on where you practise and who your callers are, so ask a qualified adviser rather than relying on a vendor's template.

Keep the transcripts useful, too. A weekly 20-minute review of ten random calls tells you more than any dashboard: you hear where it stumbled, where callers got confused and which phrases the script should change.

Mistakes that show up in the first month

The helpful-sounding advice. A tool told to "reassure callers" tells a parent that a loose bracket is "nothing to worry about" while the child has a wire in the cheek. The fix is a ban list: no advice, no reassurance about symptoms, no guesses about timing.

The silent miss. A handoff text goes to a phone that is on silent or a person on leave. Send to two named people and a shared inbox, and write down who is on duty each day.

The wrong consult length. The tool books a 15-minute adjustment slot for a first consult because the slot types are named alike. Give the new-patient type an unmistakable name and show only that one to new callers.

If you would like help writing the scripts and the test plan for your own call types, that is what my AI implementation consultation covers.

Orthodontic AI receptionist questions

Should an AI receptionist tell patients what to do about a broken bracket?

No. Let it take the details and hand the call or message to staff, who give advice. A loose bracket, a poking wire and a facial injury can all sound alike on the phone, and the receptionist should not judge which is which. Script it to collect, escalate and log, nothing more.

Can it book new-patient consultations on its own?

Yes, if it can see the right appointment type and your orthodontic software lets it write the booking. Check with a live test call, because the write-back differs by practice system and version. Start with one consult slot type, one provider group and a fixed script.

How much does an orthodontic AI receptionist cost?

Orthia, a vendor built for orthodontic practices, lists Standard from $399 a month per clinic (after-hours and overflow calls) and Premium from $499 (full 24-hour cover), month to month with no per-call fees. Other vendors price by minute or plan. Always ask what a busy month would cost.

What about practices with several sites?

Give each site its own hours, address, parking note, consult slot types and escalation number, and have the tool announce the site before it books. Test every site on a different day, because a script that works for the first location can send callers to the wrong chair at the second.

Do I need to tell callers it is an AI?

Many places expect disclosure and recording notices, and rules differ. Put a clear opening line in the script, such as saying it is the practice's automated assistant, and ask a qualified adviser what your area requires for call recording and consent.

Further reads

• AI Receptionist vs Front Desk Hire: A Dental Practice Comparison — Compares an AI receptionist with hiring a front-desk person.
• After-Hours Dental Answering Service: Can AI Take Emergency Calls? — Emergency calls when the practice is closed.
• AI Voice Agent for a Dentist: Minutes, BAA and Exit Terms to Negotiate — Minutes, BAA and exit terms to negotiate.
• What Does an AI Receptionist Cost a Dental Practice in 2026? — What an AI receptionist costs a dental practice.
• How Dental Practices Use AI for Recalls and Appointment Reminders — Recalls and reminders that keep adjustment visits on time.
• Setting Up an AI Answering Service That Never Misses an Emergency — Set up an answering service that never misses an emergency.
• What Is an AI Receptionist and How Does It Handle Bookings? — The basics of how an AI receptionist books appointments.
• Marketing for Orthodontic Practices: AI for Referrers and Parents — Add two audiences to your orthodontic marketing, referring dentists and parents, with AI drafts that pass a claims and consent check.
• Orthodontic Payment Plan Template: Fee Scripts Drafted With AI — A fillable orthodontic payment plan layout, three parent fee scripts and the checks that stop AI drafts quoting the wrong numbers.
• Orthodontic Referral Form Template and an AI Completeness Check — A field-by-field orthodontic referral form, a radiograph and consent checklist and a rule-based AI check for missing details.
• Orthodontic Staff Training: AI Role-Play Scripts for Phone Calls — Role-play prompts, a call-type list, a scoring sheet and a quiz bank for orthodontic front-desk phone training, built with AI and reviewed by a coordinator.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Orthia home, pricing and integrations pages; Weave AI Receptionist page (checked 4 November 2026). Vendor statements are treated as claims to verify in a live test.