A Telegram assistant for visa appointments
with passport OCR and browser automation
Prototypes of a Telegram assistant for visa appointments: browser automation in a headful container with remote viewing, passport OCR through a vision model, human-like interaction patterns, a proxy layer, and a unified bot that serves two visa centres with per-user modes and a shared OCR module. Built with a two-developer protocol and a changelog discipline.
What it shows
Browser automation that behaves like a person, document OCR with a vision model instead of brittle templates, and a bot that serves two different services with one code base. Listed as a prototype: the full booking flow was not completed, and we publish only what works.