The complete guide to an AI receptionist for dental practices.

Most dental practices already know reception is the bottleneck. What they don't know is how much of it can be answered correctly, on brand, at 11pm — without a single new hire. This is the playbook the practice managers we work with use, in the order they use it.

The job at the front desk hasn't changed — but the volume has. A typical four-surgery practice sees several hundred reception interactions a week: phone, web chat, WhatsApp, and email. About two-thirds of those are the same six questions repeated: new-patient pricing, NHS availability, opening hours, parking, treatment ranges, and "do you do Invisalign / implants / whitening." A modern AI receptionist clears that bottom layer so your team can focus on booking and treating patients.

The distinction we make: this isn't a marketing chatbot. It's a trained receptionist agent that has been taught your fee schedule, your clinicians, your booking rules, and your tone — and is graded on whether it gives an answer a human receptionist would have given.

We thought we needed a fifth receptionist. We needed the existing four to stop answering the same six questions all morning.Dr Helena Park · Bay Smile Studio, Brighton

A good test: an AI receptionist should be able to take any question a new patient might ask on your website and answer it the way your best receptionist would on a quiet morning. In practice that breaks into four jobs:

• Answer. Pricing ranges, hours, treatments, what's available on NHS vs private, what's in scope at each chair, how follow-ups work.
• Triage. Recognise emergencies — a lost crown, swelling, trauma — and route them to the practice line or 111 with the right script.
• Capture. Take a name, number, preferred clinician, and reason — and drop it into your practice management software or a daily digest.
• Hand off. Recognise when a human is needed (complex treatment plan, finance enquiry, complaint) and pass the conversation with full context.

Setup is one afternoon of paperwork, not a six-month integration. We ingest three things: your website, your treatment fee sheet, and a short Q&A interview with the practice manager. The Q&A is the part that matters — it's where the receptionist learns *your* tone (warm, NHS, mixed, premium private) and your booking rules ("never promise same-day with Dr Singh on a Tuesday").

What we ingest, what we never store

Many dental enquiries we see arrive between 6pm and 8am — when the practice is closed. The pattern is consistent: someone breaks a crown over dinner, panics, googles two clinics, and chooses the one that responds first. If your competitor replies at 9am the next day and you replied at 7:43pm with a price range and a Wednesday slot, you win.

This is the easiest ROI to measure: count enquiries received outside hours, multiply by your new-patient conversion rate, and divide by your average patient lifetime value. The practices we work with usually see that number recover the entire Adoomi subscription inside the first month.

The handoff is the moment the receptionist becomes trustworthy. Adoomi hands a patient over to a human in four situations, each tunable per practice:

1. Clinical risk. Trauma, swelling, severe pain, paediatric emergencies, anything the patient describes as "urgent."
2. Finance or treatment-plan questions beyond the published fee schedule.
3. Complaints or dissatisfaction. Any negative sentiment over a threshold goes to the practice manager directly.
4. Explicit ask. The patient says "can someone call me" or "I'd rather speak to a person."

This is the bit most bots get wrong. The default behaviour is not to guess. If the agent doesn't have a confident answer it says so, captures the contact, and books a callback — the same thing a good receptionist would do. The unanswered questions then surface in a weekly digest so the practice can decide whether to add them to the knowledge base.

Over a 90-day pilot, the typical practice sees the "couldn't answer" rate drop substantially, simply because the Q&A grows naturally with each unknown that's been triaged.

Practices typically start on a 60-day pilot at a flat monthly rate (no per-message billing — predictable for procurement). The rollout follows a fixed shape: kickoff day 1, content review day 3, internal-only on day 7, public-facing on day 14, full hand-off including out-of-hours on day 21.

For multi-site groups we add a clinical-governance review on day 10 before going public anywhere. See the full rollout plan →