Immigration services · Bratislava
AI sales manager
Replies to clients from the owner's real Telegram account — mirrors his writing style, his timing and the "typing…" indicator. Clients never realise they're chatting with an AI, not a real person.
Problem
Bohdan is an immigration consultant in Slovakia, handling temporary protection (€350/person) and business services for Ukrainians (€200 paid consultations, residency, s.r.o., živnosť). Inbound on Telegram was steady, but quality suffered: evening messages sat until morning, weekends ate leads, and recurring 'what documents?' / 'how long does it take?' questions soaked up more time than the actual paying clients.
Solution
I built a Telegram userbot on Telethon + GPT-4o that runs from Bohdan's real phone number — not a @bot, a live account. It imitates his messaging style and timing (2–12 minute unread windows on first contact, a typing indicator, an explicit read-receipt delay), distinguishes services (€350 temporary protection vs €200 consultation), pulls the live EUR→UAH rate from Monobank's public API, sends the right payment details (IBAN or Mono card with the converted hryvnia amount), flags payments with a [PAYMENT_SENT] marker and quietly hands the conversation over to Bohdan. Document uploads and escalations ping the owner instantly with a deep link into the chat.
Results
- ≈ 210 hours of owner time returned per month — the equivalent of 1.3 full-time staff
- Full 24/7 coverage — no more lost evening or weekend leads
- Payment details land before the prospect can DM a competitor
- Run cost ≈ $30/month in API — vs ≥ €1,200/month for a junior sales hire
“The bot writes the way I do — clients have no idea it isn't me. I open Telegram in the morning and three payments are already in, plus three notifications in the owner chat.”
Virtual employee
One digital hire instead of a whole shift
AI-Bohdan works 24/7 from a real Telegram account, replies in the owner's voice, and closes payments before the prospect has time to message a competitor.
Live dialog
From first 'hello' to payment details delivered
The bot writes like Bohdan does — with pauses, delays, and a real typing indicator. On the right, the engine events firing behind the scenes.
What the system is doing behind the scenes — in parallel with the friendly chat.
- Waiting for the first client message…
Need something similar?
30 minutes on a discovery call, and 24 hours later you have a fixed estimate.