kulov.
All work

Redesign · AI · n8n

Dr. Kulova

Site redesign migrated off Tilda, AI-structured patient messaging, and content automation on n8n

drkulova.ru

The problem

The old site ran on Tilda with a built-in CRM — a long path to a lead for the client, and limited room for content and patient-messaging automation.

What's implemented

  • Full site redesign migrated off Tilda — entry into a consultation via a "What brings you here?" block instead of a standard form
  • Receiving and structuring patient conversations on Telegram: after a pause in the dialogue, AI reads the whole thing and creates a patient card in Google Docs (complaints, date, doctor's conclusion, recommendations)
  • Content pipeline on n8n: she writes a raw note, AI turns it into a polished post in her own voice (trained on her real past posts), published to Telegram and VK after her approval
  • Two-tier content moderation — general topics get auto-published after a streak of approvals with no doctor involvement; medical topics always stay manually approved
  • Fully self-owned infrastructure: own VPS, Docker, Caddy, and n8n on a dedicated subdomain, running the entire system for about $30/month total
  • Replaces what would otherwise be a dedicated admin assistant (~$2,000/month) — she no longer manually tracks conversations, drafts posts, or keeps a patient log by hand

Result

Within 3 months of the redesign, the site drove +$3,000 in new patient leads. The automation layer does the work a ~$2,000/month admin assistant would otherwise have to: patient conversations turn themselves into structured records, marketing posts draft themselves from her raw notes, and the whole system runs on roughly $30/month of infrastructure instead of a salary.

Stack

n8nTelegram Bot APIVK APIGoogle Docs APIOpenAI API

Screenshots

Doctor's homepage with a "What brings you here?" block instead of a standard booking form

Site redesign

Instead of a "leave a request" form — a "What brings you here?" block: the visitor picks their scenario right away instead of guessing which button they need.

Patient reviews block with real 5.0 and 4.8 ratings pulled from external platforms, and live testimonials

Trust from outside platforms

Ratings and reviews are pulled from real external platforms (ProDoctorov, SberHealth), not invented for the design — another example of "every claim is checkable."