İrmet Hospital: Multilingual AI Patient Intake System

Greet international patient inquiries arriving from different time zones in each person’s own language, regardless of coordinators’ working hours.

The Starting Point

• An 8–10 hour response gap at night and on weekends
• Roughly 5 inquiries a night left unanswered until morning
• Coordinator capacity falling short in peak periods
• Inquiries moved into the CRM by hand

Patient Intake on WhatsApp

We built a text-based AI assistant that runs on WhatsApp. It greets each inquiry in the person’s own language and gathers administrative details, such as the treatment area of interest, country, language and preferred contact time, in natural conversation. It makes no medical assessments and hands the conversation over to the clinical team.

Core Components

• Greetings in five languages: Turkish, English, Russian, Arabic and German
• Administrative pre-intake gathered in natural conversation
• Handoff of each inquiry to a coordinator
• A coordinator dashboard for tracking the inquiry flow

Measured on the Dashboard During the Pilot

• 3,172 patient inquiries handled
• 2.4-second average first response time
• 1,742 natural conversations
• 491 inquiries handed to coordinators
• Inquiries from more than 10 countries, including the US, New Zealand, Ireland, Italy, Malta and the UK

Figures cover the pilot period (as of August 2026). The system gathers information and prioritizes it for coordinators; clinical decisions remain with physicians.