All case studies Growth · bots · 2025

Telegram lead bot wired into the CRM

Inbound Telegram messages go through classification and entity extraction, get scored for priority, and arrive in the CRM as complete lead records. The sales team opens a queue, not an inbox.

auto-processed
90%
instead of hours
Seconds
manual entry
Zero

The problem

Inbound enquiries arrived in Telegram as free text at any hour. Somebody had to read each one, work out what was being asked, decide whether it was worth a call, and retype it into the CRM.

How it works

A webhook handler ingests every message and hands it to a job queue, so a slow model call never blocks the bot.

Classification, intent detection and entity extraction run on the message body. Name, request type, budget, urgency and contact details come out as structured fields, get scored for priority, and are mapped onto the CRM schema by a dedicated data mapper.

The sales team opens a prioritised queue of complete lead records instead of an inbox of raw chat.

What it bought

About ninety per cent of enquiries are processed without a human touching them, the turnaround dropped from hours to seconds, and manual data entry went away.