Patient Intake Chatbot for a Denture and Implant Clinic Network
Let's Connect
Overview
What we built
A denture and implant clinic network's phones went quiet at 5pm while its older patients kept calling. We built a chatbot that answers them, day and night.
In plain terms: the network's 35 clinics serve mostly older adults, and those patients tend to ring with questions about insurance, pricing and financing rather than fill in web forms. Anything that arrived after 5pm went to voicemail, fewer than half of those callers were ever reached again, and front-desk teams began every morning working through the overnight backlog instead of greeting the patients standing in front of them.
We built a patient intake chatbot for the network's website and SMS line. It walks people through booking a consultation with simple guided choices, answers insurance and denture-care questions from an approved clinical library, and writes bookings straight into the scheduling system. Anything complex or sensitive is packaged up for staff each morning. Today 41% of all patient enquiries are handled entirely outside staffed hours, the first response to an enquiry has dropped from 9 working hours to under 1 minute, and after-hours conversations have grown to 27% of new-consultation bookings within 6 months.
The Problem
After-hours calls going unanswered
The network had grown to 35 clinics, and its enquiry pattern reflected its patients. Prospective denture and implant patients are mostly older adults who prefer to pick up the phone, and their questions, insurance cover, likely pricing, financing options, need a considered answer before they will commit to a consultation. Those calls did not stop when the front desk went home.
Everything after 5pm went to voicemail. Fewer than half of those callers were ever reached again, which meant a steady share of motivated patients quietly gave up or looked elsewhere. The callers the clinics did reach often waited the better part of a working day, since front desks spent every morning working through the overnight backlog before the day's walk-ins and check-ins took over.
The morning backlog had its own cost. Front-desk staff started each day on the back foot, returning voicemails between arrivals, and the patients standing at the desk competed with the patients on hold. Nobody could say how many enquiries were being lost, only that the voicemail box was full again by the next morning.
After-hours silence
The clinics' staffed hours ended at 5pm, but patient enquiries did not; everything that arrived later sat in voicemail until the next working morning.
Callers lost forever
Fewer than half of the patients who left an overnight message were ever reached again, and each one was a consultation that never happened.
Morning backlog ritual
Front desks began every day working through the overnight queue, so callbacks competed with arriving patients and first responses stretched to 9 working hours.
Conversations, not voicemails
Older patients rang to talk through insurance, pricing and financing before committing, conversations a voicemail greeting could not hold and a busy front desk could not always give.
What it was costing them
The arithmetic was unforgiving. With enquiries answered in 9 working hours on average and fewer than half of overnight callers ever reached, a meaningful share of prospective denture and implant patients simply disappeared. Front-desk time went into chasing voicemails rather than caring for patients in the clinic, and 35 clinics repeated the same wasteful morning routine every single day.
The Solution
Guided patient intake chatbot
We built the chatbot around how the network's patients actually behave. Rather than an open-ended chat window, the assistant leads with guided quick-reply flows: large, clear choices for booking a consultation, asking about insurance or getting denture-care guidance, so an older patient never faces a blank box wondering what to type. The same flows run on the website and on the SMS line, meeting patients on whichever channel they already use.
Behind the guided flows sits a language model with deliberately narrow permissions. It answers only from an approved clinical FAQ library, so every response reflects wording the clinical team has signed off; anything outside that library is declined and set aside for a person. Confirmed bookings write straight into the network's scheduling system rather than into a list for retyping.
Complex or sensitive queries follow a different path. The chatbot captures the details, packages the conversation with its context, and queues it for staff follow-up each morning, so front desks start the day with an organised worklist instead of a voicemail backlog. The result is a clear division of labour: the chatbot handles the routine and the repetitive, and staff handle the conversations that genuinely need a person.
Key decisions
Guided flows before free text
Quick-reply choices lead every conversation, because the network's older patients respond far better to clear options than to an empty chat box waiting for typed questions.
Answers only from approved content
The LLM draws exclusively on the approved clinical FAQ library, so no patient ever receives improvised clinical or pricing information from the chatbot.
Bookings write straight through
Confirmed consultation bookings write directly into the scheduling system in real time, removing the manual retyping step where details had previously been lost or delayed.
SMS as a first-class channel
The same intake flows run over the SMS line as on the website, because many of the network's patients are more comfortable texting than installing anything.
Mornings for follow-up, not backlog
Complex and sensitive queries are packaged overnight into a structured worklist, turning the front desk's chaotic voicemail hour into a short, prioritised follow-up routine.
Measurable Impact
What changed after launch
The clinics' day now starts differently: 41% of all patient enquiries are handled entirely outside staffed hours, and the average first response to an enquiry has fallen from 9 working hours to under 1 minute. After-hours conversations grew to 27% of total new-consultation bookings within 6 months, evidence that the patients who used to vanish into voicemail are now booking.
Front desks feel the change most. Inbound call volume has dropped by roughly a third across the 35-clinic network, and the overnight voicemail backlog has been replaced by a packaged worklist of the genuinely complex cases. Staff spend their mornings on the patients in the clinic rather than on message triage, and the network finally has a consistent record of what patients are asking.
After-hours enquiries
Voicemail, with fewer than half of callers ever reached
41% of all enquiries handled entirely outside staffed hours
First response
An average of 9 working hours to any reply
Under 1 minute, on website chat and SMS alike
Front-desk mornings
Working through the overnight voicemail backlog
A packaged follow-up worklist of complex queries only
New bookings
Dependent entirely on staffed phone hours
After-hours conversations drive 27% of new-consultation bookings
Headline results
41% of all patient enquiries now handled entirely outside staffed hours
After-hours conversations grew to 27% of total new-consultation bookings within 6 months
Average first response to an enquiry dropped from 9 working hours to under 1 minute
Front-desk inbound call volume reduced by roughly a third across the 35-clinic network
Tech & Tools Used
What powered the build
Every tool below earned its place in this engagement. Here is the part each one played.
Next.js
Renders the website chat experience, serving the guided quick-reply flows inside the network's existing site so patients start a conversation without installing anything.
Node.js
Runs the conversation service behind both channels, applying the flow logic, calling the language model and coordinating writes to the scheduling and follow-up systems.
OpenAI API
Generates answers to insurance and denture-care questions, constrained to the approved clinical FAQ library so responses stay within clinically signed-off wording.
LangChain
Manages retrieval from the FAQ library and enforces the guardrails that keep the model answering from approved content or handing off to staff.
PostgreSQL
Stores conversation records, intake details and booking outcomes, giving the network one consistent history of what each prospective patient asked and when.
Redis
Holds live session state so a patient can move between quick-reply steps, or pause and resume over SMS, without losing their place.
Twilio SMS
Carries the text-message channel, delivering the same guided intake flows to patients who prefer texting to using the website.
AWS Lambda
Executes the event-driven pieces of the pipeline, including the overnight job that packages complex and sensitive queries into the morning follow-up worklist.
HubSpot CRM API
Syncs enquiry details and follow-up status into the network's CRM so front-desk teams pick up each conversation with its full context.
Ready to Build your Dental Healthcare Business with Chatbots & Virtual Assistants
Ask Byte
Ask Byte
Typically replies instantly
just Now
Hi! I'm OrganByte's assistant. How can I help you today?
AI-generated content may be incorrect
