Case study: a Telegram assistant for a private practitioner

Context

A detailed first consultation runs tens of minutes per patient: history-taking, explaining the cause of the condition, recommending tests. On any inbound flow, the doctor's expertise hits a hard time limit.

Task

Build the doctor's digital twin — a Telegram assistant that autonomously runs the dialogue with a patient, collects symptomatology, and produces a personalized review in the style and logic of the specific specialist.

Implementation

• RAG over clinical guidelines, medical literature, and the doctor's personal materials — responses only from approved sources
• Guided dialogue: the bot actively asks questions and collects symptomatology on its own
• Personalized review across 15 blocks: causes, test recommendations, mechanism explanations, lifestyle, escalation, etc.
• Fully autonomous flow — no doctor in the loop per conversation
• Deployed as a Telegram bot

The style and approach reproduce the logic of a specific specialist rather than an averaged model.

Results

• Every patient receives a structured review against a unified standard of the doctor
• First-touch load is removed — the doctor steps in where their expertise is actually needed
• Expertise scales to a larger patient flow without quality loss

The service does not replace a doctor and does not diagnose. The final medical decision remains with the specialist.