A pediatric dental clinic in Kyiv, 4 doctors: checkups 70% → 92% in 5 months

This is a private pediatric dental clinic in Kyiv: 4 pediatric dentists, a separate play area, focused on treating baby and permanent teeth, fissure sealing, and treatment under sedation for the youngest. The audience here is special: the decision to visit is made not by the patient but by the mom or dad, and the clinic runs all its communication with the parents — from the first question to booking and reminders.

Channels: Instagram (parents actively scroll the feed in the evening, after putting the child to bed), the clinic's Telegram chat, and daytime calls. Before MTDK ai, one administrator held all bookings together during work hours. There was no patient-booking automation at all — the schedule, reminders and answers to parents' questions lived in her memory and notebook.

A key thing that sets this case apart from adult dentistry: there are no long implant deals or deposits here, but there is a cycle you'll find nowhere else — a checkup every 6 months for every child. A child's tooth grows and changes fast, and a missed half-year checkup isn't a "forgotten procedure" but a real cavity risk that parents only notice at the pain stage. So recall in pediatric dentistry isn't marketing — it's a medical necessity.

The baseline we recorded before launch: about 70% of families came back for the 6-month checkup, the other 30% simply got lost — no one reminded them. The Google rating sat at 4.4, while the nearest competitors in the area had 4.8. Many Instagram inquiries arrived after 20:00, when the administrator was no longer answering.

The takeaway from the context is simple: in a pediatric clinic the bottleneck isn't chairs or doctors but systematic communication with parents. That's exactly where we aimed the dental automation — not at booking as such, but at keeping the family on the checkup schedule.

Every pain in this case boils down to one idea: a child shows up only when an adult remembers them. And the adult is busy, with no one to remind them. We isolated three pains, and none of them repeats the adult dentistry cases.

Pain one — broken checkup recall. Parents sincerely plan to bring the child in "in six months", but six months later their head is full of a hundred other things. 30% of families simply didn't return for the checkup because the reminder didn't exist as a process. For pediatric dentistry that meant both lost revenue and — more importantly — a missed cavity later treated painfully and at greater cost.

Pain two — parents book in the evening, and there's no reply. A mom opens Instagram at 21:30, when the child has finally fallen asleep, writes "when can I bring my son for a checkup?" — and waits until morning. By morning the impulse fades, the child "isn't really complaining", and the visit slips to a vague "later". The clinic was losing exactly the warmest, most booking-ready inquiries — the ones arriving in the evening window that no administrator was physically covering.

Pain three — parents' questions ate the administrator's day. "From what age do you see kids?", "Can I bring two at once?", "My child is scared — how do you handle that?", "How long does a checkup take?" — that's not booking, it's pre-booking full of parental anxiety. While the administrator reassured one set of parents in chat on a loop, others didn't wait for a reply and left for competitors with a higher rating.

The heart of the solution is an AI clinic administrator built on Claude, tuned specifically for talking with parents. The key difference from an ordinary dental chatbot is that it holds not a "contact" but a family: it ties the name and age of the specific child to every booking. Thanks to that, all later reminders sound not like a broadcast but like personal care: "time to bring Sofia for her checkup".

Channels — a Telegram bot and Instagram Direct via the official Meta API, that is exactly where parents already sit in the evening. The AI replies instantly and around the clock. In the dialog it always clarifies two things — the child's age and the complaint (checkup, a tooth hurts, a chip, a bite consultation) — and only then offers a slot with the right doctor. If the case is sensitive (the child is scared, sedation is needed, acute pain) — the AI escalates to the administrator with full context, so the mom doesn't have to retell everything a second time.

The case's main lever is automatic checkup recall. After every visit the AI sets a "6-month timer" for the child. A week before the control date it writes the mom a personal reminder with the child's name and an offer to pick a convenient evening slot right away. This exact scenario, built on n8n as the schedule orchestrator, turned one-off visits into a regular cycle and pulled checkups from 70% to 92%.

Recall is complemented by short reminders about an already booked visit — 1 day and 2 hours before, also mentioning the child's name. For pediatric dentistry this is critical: parents often mix up days because of kindergarten, clubs and illnesses, and a gentle reminder the day before removes half the no-shows before they happen.

We separately defined the tone: warm, calm, parent-to-parent, without bureaucratic phrasing and without excessive emojis. The AI never gives medical advice about treating a child — any question about a symptom it converts into a booking for an in-person exam. In a medical niche, and a pediatric one at that, these boundaries are the basis of trust.

One screen is worth a hundred explanations. Below are two scenarios in a row: first the evening booking, when a mom writes at half past nine in the evening and the AI clarifies the child's age and complaint and immediately offers a slot; and then the same recall six months later, where the reminder already arrives with the specific child's name.

Notice: no "an administrator will reply tomorrow". At 21:30 the mom gets a concrete time, and six months later the clinic reminds her about Sofia itself — and the family stays on the checkup schedule even if the parents forgot about it. That's how patient-booking automation in the pediatric niche turns into retention rather than just a fast reply.

Days 1–3. Data gathering for the pediatric specifics: the schedule of 4 doctors, checkup duration by age group, the price list, and the list of complaints kids are most often booked for. A separate interview with the administrator — which parent questions repeat daily and what exactly she answers word for word, so the AI speaks in her voice.

Days 4–6. Setting up the AI and the recall logic in n8n: every completed visit is assigned a next-checkup date in 6 months, tied to the child's name and age. Testing in an internal Telegram chat — about 40 dialogs where we deliberately break the scenarios: an anxious mom, two children at once, acute pain at night.

Days 7–9. Launch on the clinic's Telegram channel — with a loyal audience that forgives small awkwardness. In parallel we monitor every dialog and tune the prompt to parents' real phrasing.

Days 10–12. We connect Instagram Direct — the largest and warmest evening channel. For the first days we run in "AI replies, a human sees every dialog" mode, because the cost of a mistake here is highest.

Day 12+. Full autonomy in booking and recall. The administrator moves to lounge work and complex cases, and the first recall messages start bringing back the parents who had long "meant to come but forgot".

Five months after launch we tallied the numbers. These are data from the clinic schedule, the bot statistics and the Google profile, not marketing roundings:

Inquiries: +35%. The main growth came not from the daytime phone but from evening Instagram and Telegram inquiries that used to go unanswered until morning and mostly faded away.

Checkups: 70% → 92%. This is the headline number of the case. The automatic 6-month recall tied to the specific child brought back into the schedule the third of families that used to get lost. For pediatric dentistry that's not only revenue — it's less neglected decay caught at the pain stage.

Google rating: 4.4 → 4.8 in 5 months. It rose not as a separate goal but as a consequence: when parents get answers within seconds in the evening and a gentle reminder about their child, they more readily leave a warm review. The rating here is a side effect of well-tuned communication, not the subject of the case.

Parent behavior changed qualitatively: they now book their kids in the evening with no calls at all. The warmest window — from 20:00 to 23:00, when the child is already asleep and the adult finally has a minute — stopped being "dead" and became the main channel for booking checkups.

We'll also note the effect on the administrator: parents' routine questions ("from what age", "can I bring two children") are now handled instantly by the AI, so the person focuses on the kids in the lounge and the anxious parents who genuinely need live contact.

No magic here: the live administrator hasn't gone anywhere — the AI only took the evening shifts and systematic reminders so the human could focus where a pediatric clinic needs live contact. Here's how the key points of working with parents changed before and after launch:

For the first days the recall sent a dry broadcast "a reminder about your checkup" with no child's name — and parents perceived it as clinic spam. We fixed it: now every reminder includes the specific child's name and gentle phrasing, so it reads as care, not a mass mailing.

At first the AI offered all free slots in a row, and an anxious mom got lost among twenty options. We did it differently: the bot offers the 2–3 nearest evening slots (since evenings are exactly what suits parents), asks "does this work", and only shows more after a refusal.

At launch the AI tried to answer questions about a child's symptoms ("if the gums turned red, it could be…"). For a pediatric medical niche this is unacceptable. We wrote a hard rule: any complaint about a child's condition → "a doctor needs to examine this, I can book you the nearest time".

The recall initially counted 6 months from the booking date, not from the actual visit date — so when a visit was rescheduled, the reminder arrived off time. We tied the timer to the completed visit in the schedule. The takeaway is simple: in pediatric recall the reference point must be the real exam, not the intention to come.

Now the main question of any owner — how much it costs and when it pays off. For this clinic we used the mid-tier package: a monthly service of €150/mo (≈UAH 7,700) plus a one-time setup of €600 (≈UAH 30,800). Two integrations (Telegram and Instagram) and the recall logic in n8n fit exactly into the mid level — without the heavy medical integrations that adult chains need.

Where the additional revenue comes from: recovered checkups and evening inquiries that didn't exist before. The third of families that used to get lost returned to the regular six-month cycle — and every checkup pulls along sealing, treatment and the next visit. Plus +35% new inquiries from the evening window, which cost the clinic zero extra ad budget.

Payback of the full package (service + setup in the first month) — about one month. From the second month on, the monthly fee pays for itself with a few recovered checkups. For comparison: keeping an administrator on evening shifts and weekends at a pediatric clinic is economically unjustified, while doing recall by hand simply doesn't work systematically — a human forgets exactly the way the parents do.

As the owner put it: "we thought we had a booking problem, but it turned out to be a problem remembering the child". When families disappear after the first visit and evening inquiries go unanswered, it's not about staff but about no one recalling a specific child on time. That's exactly what the automation takes on — quietly, every day.