Discuss a project
ServicesCasesTechnologyBlogAboutContact Discuss a project
ENRUDEESFRITPLUAUZKZCZPTNLETLVLT
UZBEKISTAN · TASHKENTHEALTHCARE · MULTI-SPECIALTY CLINICAI AUTOMATION · CHATBOTTELEGRAM · WEBSITE2025–2026

Medical clinic: 65% of bookings automated and 520 fewer calls every week

65%
OF BOOKINGS AUTOMATED
−28%
NO-SHOW RATE
−520
RECEPTION CALLS PER WEEK
2
LANGUAGES: UZBEK & RUSSIAN
Multi-specialty clinic: 65% of bookings automated
PROBLEM

Too many calls, too many missed appointments, too little front-desk capacity

The clinic’s reception team was spending most of the week answering the same 6–8 questions: doctor availability, branch, price, preparation and rescheduling. Patients increasingly preferred messaging in Uzbek or Russian, especially via Telegram, but the clinic had no self-service flow connected to the live schedule.

✕Reception handled 520 repetitive calls per week that could be automated;
✕Booking depended on manual phone coordination with the MIS schedule;
✕No-show rate was inflated by weak reminder and confirmation workflows;
✕No bilingual self-service in Uzbek and Russian on Telegram or the website;
SOLUTION · 8 WEEKS

A bilingual AI bot connected to the real schedule and built for conversion

We designed one booking logic for two channels — Telegram and website chat — with live access to doctor slots in the clinic’s MIS. The bot answered routine questions, offered available times, confirmed appointments, and triggered reminder sequences that reduced forgetfulness and manual follow-up.

WEEKS 1–2
Discovery and booking logic
Mapped 40+ reception scenarios, built Uzbek/Russian intents, and defined flows for new booking, rescheduling, cancellation, branch selection and specialty routing.
WEEKS 3–4
MIS integration
Connected the bot to the clinic schedule, doctor availability and appointment statuses so patients could see real slots instead of leaving callback requests.
WEEKS 5–6
Telegram + website launch
Deployed the assistant in Telegram and on the website, with fallback to reception for edge cases, lead capture, and language switching in 1 tap.
WEEKS 7–8
Reminder automation and tuning
Added confirmation, reminder and reactivation messages 24 hours and 2 hours before visits; optimized prompts weekly using drop-off and handoff data.
RESULT · 3 MONTHS

The bot became the clinic’s main booking layer for routine demand

METRICBEFOREAFTERCHANGE
Bookings handled automatically 0% 65% NEW CHANNEL
Reception calls / week 800 280 −520
No-show rate 18% 13% −28%
Average booking time 7.4 min 2.1 min −72%
Share of bookings via Telegram 0% 49% NEW CHANNEL
“We expected a chatbot. What we got was a second reception desk that works 24/7 in two languages and never loses the schedule.”
Operations Director MEDICAL CLINIC · TASHKENT
FAQ

Frequent questions

Didn’t find your answer — ask on Telegram, we reply within 2 hours.
How long does it take to launch a bilingual booking bot like this one?
This rollout covered discovery and scenario mapping, live schedule integration with the clinic's MIS, and deployment across Telegram and the website, with weekly tuning afterward. Within that process the clinic reached 65% automated bookings and cut reception calls by 520 per week.
Would this work for a clinic that currently only books appointments by phone?
Yes, that was this Tashkent clinic's exact starting point, with booking depending entirely on manual phone coordination against the clinic's schedule. Connecting a bilingual bot to that same live schedule let patients self-book in Uzbek or Russian, automating 65% of bookings without removing the phone option for edge cases.
What was the hardest part of building a bot that handles bookings in two languages?
Mapping the logic was harder than the translation itself: over 40 reception scenarios, covering new bookings, rescheduling, cancellations, and specialty routing, had to be defined for both Uzbek and Russian and connected to live MIS data. That groundwork is what let the bot handle 65% of bookings unattended.
How do you measure whether a booking bot is actually working?
The core measures were automation share, call volume, and no-shows. Reception calls dropped from 800 to 280 per week, no-show rate fell from 18% to 13% after adding 24-hour and 2-hour reminders, and average booking time dropped from 7.4 minutes to 2.1 minutes.
NEXT CASE Freight company: quote requests ×2.8, top-3 in 4 countries
Want the same result? Free audit of your website and ads — within 48 hours. Get an audit →