Answering service or AI front desk?
The short answer
Pick a human answering service if most of your missed calls need judgement — upset patients, complex insurance, clinical triage. Pick an AI front desk if most of your missed calls are routine booking and rescheduling, which for the majority of clinics is where the volume actually sits. The real answer for most practices is both, with the AI in front and a human path behind it.
The decision
Should a clinic put a human service or an AI system on the calls it misses?
We do not publish anyone's prices or feature lists here. Vendor pricing changes monthly, a stale table is worse than no table, and we will not print a number about someone else's product that we cannot stand behind. What is below is what to judge on — ask each vendor these directly.
- 01
What happens on the calls it cannot handle
Every system fails on some calls. The one that matters is what it does at that moment — dead end, voicemail, or a warm hand-off to a person who has the context.
Watch for A demo that only shows the happy path. Ask to hear a recording of a hand-off, not a booking.
- 02
Whether it writes into your calendar or just takes a message
A message is a second job for your front desk. A booking is a patient. The difference between the two is most of the value.
Watch for "Integrates with your booking system" that turns out to mean a daily CSV, or a one-way sync that cannot see a slot someone just took.
- 03
What it does with a caller who is upset or in pain
A dental emergency and a cancellation sound nothing alike, and getting this wrong costs you the patient permanently, not just the appointment.
Watch for A system with no escalation rule at all. Ask the vendor to describe the exact trigger, not the philosophy.
- 04
Whether it identifies itself honestly
A patient who works out mid-call that the confident voice was a machine pretending to be a person does not book. Several US states now regulate disclosure directly.
Watch for Any vendor selling "indistinguishable from human" as the feature. That is the failure mode, not the benefit.
- 05
How it handles patient information
Anything that touches health information needs a business associate agreement and a defined retention window, and the vendor has to be able to hand you both.
Watch for A vendor who says "we are HIPAA compliant" and cannot produce a BAA, or who cannot tell you where recordings live and for how long.
- 06
What it costs when your call volume doubles
Per-minute and per-call pricing look cheap at demo volume and stop looking cheap in your busiest month, which is exactly when you need it most.
Watch for A quote with no worked example at your real volume. Ask for the bill at 2x and 5x your current minutes.
- 07
Who owns the phone number and the recordings
If leaving means losing your number or your call history, the switching cost is the product, not the service.
Watch for Numbers ported into the vendor's account rather than yours.
- 08
What it does after hours and on the calls nobody returns
Most of the loss is not the call itself, it is the call that was never followed up. A system that only answers is half a system.
Watch for No outbound follow-up at all — no text back on a missed call, no second attempt.
What actually goes wrong on a clinic’s phone
Before choosing between the two, it is worth being precise about what is broken, because the two options fix different halves of it.
Published call-tracking benchmarks put the share of calls to US local businesses that reach a live person at 37.8% — an independent audit of 85 businesses across 58 industries. Of the callers who do not get through, published benchmarks put the share who hang up without leaving a voicemail at 80–86%, and the share who call a competitor instead at 62%. These are industry figures, not our results.
Read that as three separate problems:
- Nobody picks up. Your front desk is with a patient, it is lunch, it is 7pm.
- Nobody calls back. The message that did get left sits in a voicemail box until the afternoon.
- Nobody follows up. The caller who hung up is invisible — you do not know they existed.
A human answering service is good at the first. An AI front desk is good at all three, but it is worse than a human on the calls that need judgement. That is the whole trade.
Where a human answering service is genuinely better
Do not let anyone sell you AI as strictly superior. It is not.
A trained human is better when the call is unusual — an angry patient, a billing dispute, an insurance question with three conditionals, a caller whose English is halting, a situation where the right answer is to stop talking and listen. Humans are also better at the thing nobody writes on a feature list: knowing when a script does not fit and abandoning it.
If your missed calls are mostly of that kind — and for some practices, particularly ones dealing with complex restorative work or heavy insurance friction, they are — put a person on them.
Where an AI front desk is genuinely better
An AI system wins on volume, hours and consistency, and on the two problems a human service usually does not touch at all.
It answers the second and third simultaneous caller, which a small human service often cannot. It is awake at 11pm when someone with a broken tooth is calling every practice in the city in order. And it does the follow-up: a text back within seconds of a missed call, a second attempt on the enquiry that went quiet, a reminder that reduces the no-show.
That last part is where most of the recovered money actually is, and it is the part clinics most often leave out of the comparison because a human service never offered it.
What neither of them fixes
Both are answering problems. Neither is a demand problem.
If your phone is not ringing, an AI front desk answers a quiet line perfectly and changes nothing. That is a visibility problem — the map pack, the profile, the treatment pages — and putting a phone system in front of it is spending money in the wrong place. The audit exists partly to tell you which of the two you have, because the answer decides where the first dollar goes.
Both also fail the same way when the calendar behind them is a mess. A system that books into a calendar nobody maintains produces double-bookings faster than a human ever could.
The configuration most clinics actually end up with
Not one or the other. AI in front, human path behind.
The AI takes the call on the first ring, handles the routine majority — booking, rescheduling, hours, directions, price ranges, sending the intake form — and hands anything outside that to a person, with the context already gathered. Out of hours, it books and texts. During hours, your own front desk is the human path, and they only see the calls that needed them.
The reason this configuration wins is not that AI is clever. It is that it changes what your front desk spends the day on: not answering, but the calls where a person actually adds something.
How to test any vendor in one call
Ask for a live demo on your own number, not a recorded one, and run these four:
- Book an appointment, then immediately call back and move it.
- Say something a system should refuse — a specific clinical question about a medication.
- Get angry, and ask for a human.
- Call after hours and say you are in pain.
Whatever the marketing said, those four calls tell you what you are buying. If a vendor will not let you run them before you sign, that is the answer.
Before you sign anything
What owners ask us
Is an AI front desk cheaper than a human answering service?
Usually per call, not always in total. Human services typically bill per minute or per call, so cost scales with volume; AI systems more often carry a flat monthly fee plus telephony. At low call volume a human service can be cheaper. Ask both for a worked example at your actual monthly minutes rather than comparing headline rates.
Will patients hate talking to AI?
Patients react badly to two specific things — a system that pretends to be human, and a system that traps them with no way to reach a person. Handled honestly, most callers care far more about being answered at all than about who answered. Anyone claiming zero friction is overselling.
Can an AI front desk answer clinical questions?
It should not, and ours does not. Answering a clinical question through an automated system is a liability and a patient-safety problem. Clinical questions route to a human. The system is there for booking, rescheduling, directions, hours, pricing ranges and follow-up.
What about HIPAA?
Any vendor handling calls that touch health information needs to sign a business associate agreement and define how long recordings and transcripts are kept. That applies to human answering services exactly as much as to AI ones. We are HIPAA-aware and build to documented practice; we are not lawyers, and a compliance officer should review whatever you choose.