Skip to main content
    All case studies
    BlueprintDental & physio clinic groups

    Dental Booking Platform With an AI Receptionist

    Online booking across clinics and chairs, reminders and recalls to cut no-shows, and an AI phone receptionist that hands off to your team.

    An example of how we'd build this for a company in this niche: the plan, the architecture and the targets we'd start from. Not a specific client's story.

    Target
    –25%
    no-show rate within six months of launch
    Target
    30%
    of new bookings made online, within six months of launch
    Target
    –50%
    unanswered calls at peak times, three months after the phone agent goes live
    The situation

    Who this is for

    Picture a regional group of 12 dental clinics with a few physiotherapy practices alongside, where front desks take hundreds of calls a day and most bookings still happen by phone. No-shows leave expensive chairs idle, six-month recalls go out late, and callers give up during the morning rush. The goal is online booking across every location, fewer empty chairs, and an AI receptionist that handles simple rebooking and overflow calls while passing anything clinical to a person.

    The hard part

    What makes it hard

    • Availability depends on more than a calendar: the right chair, clinician, room equipment and slot length for each treatment type.
    • The practice-management system stays the source of truth, but many offer limited or read-only APIs, which makes double bookings a real risk.
    • No-shows hurt most on long slots such as implant or root-canal appointments, and a generic text reminder does little to prevent them.
    • Phone lines peak at opening time and over lunch, exactly when front-desk staff are busy checking patients in.
    • Medical history and consent forms are still on paper or PDFs, and they must be current before treatment starts.
    The product

    What we'd build

    Multi-location online booking

    Patients choose a treatment, clinic and time, or book a whole series of physio sessions at once. The engine checks chair, clinician and room rules and writes each booking into the practice-management system, not a parallel calendar.

    Reminders, confirmations & waitlist

    Text and email reminders with one-tap confirm or reschedule, optional deposits for long slots, and freed slots offered automatically to patients on the waitlist.

    Recall campaigns

    Check-up and hygiene recalls based on each patient's last visit and recall interval, sent in batches the front desk can review first.

    AI phone receptionist

    Answers overflow calls and calls outside opening times, identifies the patient, and books, moves or cancels appointments. Anything clinical or unclear goes to staff with a call summary, and urgent calls follow the clinic's own emergency route.

    Digital intake & consent

    Medical history, medication, allergies and treatment consent completed on the patient's phone before the visit, then filed to the patient record.

    Group dashboard

    Chair utilisation, no-shows, recall response and call outcomes per clinic, so managers can compare locations and adjust the rules.

    Under the hood

    Architecture

    From the people who use it down to the hardware and third-party systems it talks to.

    1. Apps
      • Patient booking web app
      • Front-desk console
      • Group management dashboard
    2. Platform
      • Scheduling & rules engine
      • Reminders, recalls & waitlist
      • Forms & consent
      • Audit log
    3. Voice & AI
      • Speech-to-text & text-to-speech
      • LLM agent with booking-only tools
      • Handoff to staff with call summary
      • Call logs for review
    4. Integrations
      • Practice-management system
      • Clinic phone system (SIP)
      • SMS & email gateway
      • Payment provider for deposits
    The plan

    Delivery plan

    The same four phases as every project we run - see how we work.

    1. Weeks 1–3

      Discovery

      Shadow the front desks at two clinics, sort a few weeks of call logs by reason, and test what the practice-management system's API can actually read and write.

    2. Weeks 4–6

      Design & architecture

      Booking rules per treatment and clinic, patient flows, and the voice agent's scope: which calls it may handle and exactly when it hands off.

    3. Weeks 7–18

      Build

      Booking and reminders go live at a pilot clinic first, then recalls and intake forms. The phone agent starts on calls outside opening times before it takes daytime overflow.

    4. Weeks 19–20

      Launch & handover

      Clinic-by-clinic roll-out, front-desk training, weekly review of agent handoffs, and handover of code, prompts and runbooks.

    The team

    • Product-minded tech lead
    • Full-stack engineers
    • Voice AI engineer
    • UX/UI designer
    • QA engineer

    Compliance & security

    • Medical history and treatment details are health data under GDPR Art. 9; they stay in the practice-management system where possible, and the booking layer stores only what it needs.
    • Call recording and AI transcription need a clear notice at the start of each call and agreed retention periods; we'd confirm the lawful basis with your DPO.
    • Under the EU AI Act's transparency rules, callers should be told they're talking to an AI. A booking-only agent is unlikely to be high-risk, but we'd review the classification if its scope grows.
    • Software that assesses symptoms could count as a medical device under the EU MDR, so the agent doesn't triage or give clinical advice.

    Tech stack

    • TypeScript
    • Next.js
    • NestJS
    • PostgreSQL
    • Redis
    • SIP / WebRTC
    • Streaming speech-to-text & TTS
    • LLM with tool calling
    • OpenTelemetry
    • AWS (EU region)

    Questions we usually get

    Will it work with our practice-management system?

    It depends on what your system allows: some offer full read and write APIs, some only read access or a partner program, and a few need a sync agent on a clinic server. We'd test yours in discovery and design the booking flow around what it safely supports, rather than keeping a second calendar that drifts.

    What if the AI receptionist gets something wrong?

    Its scope is deliberately narrow: identify the patient, book, move or cancel appointments, and answer practical questions. Anything clinical, emotional or unclear goes to a person with a summary, and every call is logged so the team can review it and tighten the rules.

    Do patients have to download an app?

    No. Booking, reminders and intake forms work in the browser from a text or email link. An app is optional for groups that want one.

    More blueprints

    Building something like this?

    30 minutes with the engineers who'd build it. We'll test this plan against your situation: scope, integrations and budget.