What AI can do · Dental & medical practices

AI in Dental and Medical Practices

Almost everything written about AI in healthcare is clinical, and almost none of it is what an independent practice can use. The work that pays here is the business of the practice: whether patients say yes, whether the payer pays in full, and whether the schedule fills. Here is what is genuinely running, including the parts the vendors do not tell you.

17 things AI is doing here

01Getting patients to say yes3
02Getting paid what you are owed5
03Filling the schedule5
04Inside the visit2
05If you run more than one location2

Proven 6Working 9Early 2

Hear one working

This is the front desk, answering at nine at night.

Not a recording and not a video of one. Press call and talk to the same thing that would answer for your practice when the desk is with a patient, for a general practice we set up to show you. It books, it handles the insurance question, and it refuses everything a front desk should refuse.

Say you broke a tooth and see how fast it offers you a time.
Ask what a crown costs. It will give a range and stop there.
Ask it whether your pain sounds serious. It will not answer that.

After-hours line

BlackSig Dental

Nina, front desk

0:00

It is a demo, so the practice is made up and the line takes no patient information. Calls are capped at three minutes.

Tap any of these to see what it does and what it takes. Add the ones that sound like your business and we will show you the system they add up to.

Proven Widely used, low riskWorking Real deployments, needs setting up properlyEarly Promising, still rough at the edges

Getting patients to say yes

Case acceptance is the number that decides the month, and it is where the most interesting work is happening.

Becomes Case Acceptance System

Show the patient their own result before they decideRenders a photo of the patient’s own face with the finished smile already in it, before they agree to anything.ProvenDetails

Renders a photo of the patient’s own face with the finished smile already in it, before they agree to anything.

What changes: Case acceptance on aesthetic work. In a 2025 study of 80 patients who were consulted conventionally and then shown a simulation, 85 percent accepted the proposed treatment.
What it takes: A photo protocol the team can repeat, retracted and unretracted, consistent light, patient upright. Bad photos in, useless simulation out. A coordinator who shows the image before the fee. And consent language that says a simulation is not a clinical promise.

For exampleA patient who has been "thinking about it" for two years sees her own face with the work done, and books that afternoon.

Know which of the three you are buying: a scanner-based simulator, a selfie tool for capturing leads, or full Digital Smile Design, which is a designed case rather than a filter. A practice that buys the lead tool expecting the designed case is always disappointed. Also worth knowing: shown side by side, a skilled technician’s designs still beat AI-generated ones. The preview is what moves acceptance; AI is what makes it cheap enough to run on every patient instead of one big case a month.

Use the annotated x-ray as the conversation, not the diagnosisMarks up the radiograph so the patient can see what the dentist sees, as the thing shown during the treatment conversation.WorkingDetails

Marks up the radiograph so the patient can see what the dentist sees, as the thing shown during the treatment conversation.

What changes: Whether a patient believes the recommendation. It is a communication instrument that happens to be clinical software.
What it takes: Integration with the imaging system and a rule about who interprets what. The dentist still diagnoses.

For exampleThe patient who nods and does not book sees the highlighted fracture line on their own molar, and asks how soon you can do it.

Some owners distrust these tools because the same vendors sell to insurers as well as practices. Worth asking any vendor directly which side they are on.

Design the financing conversation as a real stepMakes the payment options part of the treatment presentation, tracked like any other step, rather than a question at the desk on the way out.WorkingDetails

Makes the payment options part of the treatment presentation, tracked like any other step, rather than a question at the desk on the way out.

What changes: Acceptance on larger cases, which is where the money is and where "I need to think about it" usually means "I do not know how to pay for this".
What it takes: Your financing options in one place, a script the coordinator follows, and tracking so you can see where cases stall.

For exampleA $6,400 treatment plan gets presented with the monthly figure alongside it, and the patient says yes at the chair instead of going home to think.

Getting paid what you are owed

The payer side is where the specialised work lives, and where most practices are quietly losing money they have already earned.

Becomes Revenue Cycle System

Catch the payer downcoding your claimsCompares what you submitted against what was paid, code by code, and flags where a payer quietly paid a cheaper code than the one you billed.WorkingDetails

Compares what you submitted against what was paid, code by code, and flags where a payer quietly paid a cheaper code than the one you billed.

What changes: Money you already earned. Most practices never see this pattern because it only shows up across hundreds of claims.
What it takes: Your claims and remittance history, and someone who will actually appeal.

For exampleYou find that one payer has been downgrading the same procedure for eight months, and appeal the lot.

The law moved in your favour here. Indiana now bars insurers from downcoding by automated system alone, and in 2026 alone more than 100 bills were introduced across 37 states, with 16 states enacting new rules. Payers are getting constrained; nothing stops you automating the detection.

Renegotiate your PPO fees with your own dataWorks out, payer by payer and code by code, how far below your local market your contracted fees sit, and turns it into the table you take into the renegotiation.WorkingDetails

Works out, payer by payer and code by code, how far below your local market your contracted fees sit, and turns it into the table you take into the renegotiation.

What changes: Your fee schedule, which most owners believe is fixed. It is not, and the leverage is data you already hold and have never assembled.
What it takes: Two to three years of claims and verification history, benchmark data for your area, and a willingness to be told no twice.

For exampleYou walk into the renegotiation with a code-by-code table showing where you sit against the local market, rather than a general complaint.

The practical order matters: negotiate the payers with the least patient volume first. Low risk if it goes badly, and a signed uplift becomes the comparable you carry into the payer that actually matters.

Audit your own perio coding, both directionsChecks charting against the codes billed, and flags both the claims that will not survive an audit and the money left on the table by defaulting to the cheaper code.WorkingDetails

Checks charting against the codes billed, and flags both the claims that will not survive an audit and the money left on the table by defaulting to the cheaper code.

What changes: Audit exposure and revenue at the same time. The periodontal codes are the most-audited error in dentistry, in both directions.
What it takes: Charting and claims in one place, and a hygiene team willing to have the conversation.

For exampleA quarter of your scaling claims turn out not to meet the definition for the code used, and you fix it before the payer finds it.

Verify benefits before the patient is in the chairChecks coverage, remaining benefit and history with the payer ahead of the visit and writes it back into the schedule.ProvenDetails

Checks coverage, remaining benefit and history with the payer ahead of the visit and writes it back into the schedule.

What changes: Hours a week on hold, and visits that start without anyone knowing what is covered.
What it takes: Portal access for every payer, and a plan for when a portal changes.

For exampleTomorrow’s eighteen patients are verified overnight, and the two with lapsed plans get a call before they arrive.

The clearest demand signal in dentistry right now: 13.6 percent of dentists already use AI for verification and another 32.6 percent plan to, the biggest planned-adoption gap of any task in the practice.

Write the narrative the claim actually needsDrafts the supporting narrative and checks the attachments are complete before the claim goes out, against that payer’s rules.WorkingDetails

Drafts the supporting narrative and checks the attachments are complete before the claim goes out, against that payer’s rules.

What changes: First-pass acceptance, and how many denials you have to work at all.
What it takes: Clearinghouse access and payer rules kept current.

For exampleThe crown claim goes out with the narrative and the pre-op image attached, instead of coming back three weeks later.

There is no benchmark denial rate worth quoting at you. Across marketplace insurers, denial rates run from 1 percent to 54 percent depending on the insurer. Only your own number means anything.

Filling the schedule

A practice with empty chairs and a full patient list has a retrieval problem, not a marketing problem.

Becomes Schedule Fill System

Mine the treatment nobody ever scheduledGoes through the diagnosed treatment that was never booked, and the hygiene patients who never reappointed, and works the list.ProvenDetails

Goes through the diagnosed treatment that was never booked, and the hygiene patients who never reappointed, and works the list.

What changes: Production from work you have already diagnosed and already sold once.
What it takes: Treatment plan and recall data that is actually accurate, which is usually the real project.

For exampleSix hundred patients with accepted but unscheduled treatment get worked over a fortnight, and forty of them book.

Question the six-month recallSets recall intervals by the patient’s actual risk rather than by habit, and schedules accordingly.EarlyDetails

Sets recall intervals by the patient’s actual risk rather than by habit, and schedules accordingly.

What changes: Chair time spent where it does something, and the number of patients you can look after properly.
What it takes: Risk data in the chart, and a clinical decision the owner has to make and stand behind.

For exampleLow-risk patients move to a longer interval and the freed hygiene time goes to the patients who actually need watching.

Worth knowing before a vendor tells you otherwise: a large randomised trial with four-year follow-up found no difference in outcomes between six-month, risk-based and twenty-four-month recall for adults. The six-month interval is convention, not evidence. Almost nobody sells tooling for this, which makes it the clearest gap in the market.

Answer the phone and actually book itAnswers around the clock and books, moves or cancels directly in the schedule.ProvenDetails

Answers around the clock and books, moves or cancels directly in the schedule.

What changes: The share of calls answered at all, and after-hours calls that become appointments instead of voicemail.
What it takes: A scheduler it can read and write, and rules for when it must hand over to a person.

For exampleA broken crown calls at 7:50pm and is booked into the 9am opening before the front desk arrives.

Find out why the caller did not bookReviews recorded calls and tells you which ones did not convert and what was said when they stopped.WorkingDetails

Reviews recorded calls and tells you which ones did not convert and what was said when they stopped.

What changes: The front desk conversation, which is the highest-value untrained skill in most practices.
What it takes: Call recording, a view of what a good call sounds like, and someone who will coach.

For exampleYou find that half the lost calls stalled on the same question about insurance, and fix the answer.

Fill tomorrow’s gap when someone cancelsWorks a waitlist the moment a slot opens and offers it to the patients most likely to take it.ProvenDetails

Works a waitlist the moment a slot opens and offers it to the patients most likely to take it.

What changes: Slots filled per week, which is immediate money rather than a projection.
What it takes: A waitlist that exists and a way to message patients.

For exampleA 2pm cancels at 11am and is filled by 11:20.

Predicting who will no-show is a much less settled idea. Filling gaps works today.

Inside the visit

The clinical side is not what we build, but two things next to it are worth knowing about.

Becomes Chairside Documentation System

Write the note while you are with the patientListens with the patient’s consent and drafts the clinical note for the clinician to review and sign.ProvenDetails

Listens with the patient’s consent and drafts the clinical note for the clinician to review and sign.

What changes: The honest reading matters here: a 2025 study found burnout fell from 51.9 percent to 38.8 percent, while documentation time fell only 8.5 percent. It makes the day feel different more than it makes it shorter.
What it takes: Consent at the start of each visit, a signed agreement with the vendor covering patient data, and a clinician who reviews every note.

For exampleYou finish at 5pm with the notes drafted, and review and sign instead of writing them at home.

Chart perio by voice, hands freeRecords the pocket depths as they are called out, so hygiene does not need a second person or a pause to type.WorkingDetails

Records the pocket depths as they are called out, so hygiene does not need a second person or a pause to type.

What changes: Whether full charting actually gets done on schedule, which decides whether the perio coding above is defensible.
What it takes: A microphone setup that works in an operatory and practice management software it can write into.

For exampleA full-mouth charting gets recorded as it is called, and hygiene runs to time.

If you run more than one location

Groups and DSOs have a question single practices do not: which location is the outlier, and why.

Becomes Multi-Site Control System

See which location is diagnosing differentlyCompares diagnosis and treatment patterns across locations and providers, and shows you where one sits well outside the rest.EarlyDetails

Compares diagnosis and treatment patterns across locations and providers, and shows you where one sits well outside the rest.

What changes: Whether you find out about a pattern from your own data or from a payer audit.
What it takes: Clinical and claims data across locations in one place, and a conversation with a provider that has to be handled well.

For exampleOne location turns out to diagnose a particular procedure at three times the rate of the others, and you look into it before anyone else does.

Stop the referrals you never hear back aboutTracks referrals out to specialists and back, and flags the ones that went nowhere.WorkingDetails

Tracks referrals out to specialists and back, and flags the ones that went nowhere.

What changes: Revenue lost to leakage, and the patients who quietly never completed their treatment.
What it takes: Referral records and a specialist network willing to close the loop.

For exampleYou see that a third of your implant referrals never came back, and find out where they went.

What you just built

Nothing picked yet.

Add anything above that sounds like your business. They group into the systems we build, so by the time you reach the form you are sending us a scope rather than a question.

HIPAA compliant

Built to work with protected health information.

Patient data is the first question a practice asks, and rightly. Here is what our compliance actually consists of, rather than a badge.

A signed BAA before anything movesWe sign a business associate agreement with your practice before any system touches patient information, and we require one from every vendor in the chain behind it.
No patient data in general-purpose toolsNothing goes into a consumer AI account. Systems that touch patient information run on infrastructure covered by the agreements needed to carry it.
Every tool in the chain is mappedThe usual failure is not the AI. It is the transcription, analytics or monitoring tool quietly sitting behind it. We list them, and you see the list.
Consent and recording designed inCall recording rules differ by state and are being litigated right now. We settle how consent works before a voice system goes live, not after.
Access, audit and retention set deliberatelyWho can see what, what gets logged, and how long anything is kept are decisions made at design time and written down, so they can be shown to anyone who asks.

Being straight with you

What AI is not doing here yet.

AI does not design a better smile than a good technician

Shown side by side in a 2025 study of 320 people, manually crafted smile designs beat AI-generated ones in every case. What AI changes is cost and speed, so you can show a preview to every patient instead of one a month. Anyone selling you better aesthetics is overstating it.

Your vendor list is a compliance list

Anything touching patient information needs a signed business associate agreement, and so does every tool sitting behind it. In April 2026 a dental front-office software vendor settled with regulators over a breach affecting 15 million people, with findings that it had never done a risk analysis. That is this exact category of software.

Recording the call is its own question

Roughly eleven states require everyone on a call to consent, and two 2025 court rulings let claims proceed against AI voice vendors for listening in. Settle this before a voice agent goes live, not after.

We do not touch the clinical side

Diagnostic AI on x-rays and images is real, regulated, and not what we build. The profession agrees on where the line sits: 82.6 percent of dentists said they would not use AI for treatment recommendations.

Where to start

Which of these is worth doing first in your business?

That is the whole question, and it is what a roadmap answers. Tell us how your business runs and we will come back with what we heard and where we would start.

vic@blacksigsystems.com

See another kind of business

Message sent.

We'll get back to you within 24 hours.