Complex apps and websites were a barrier for the general population.
A leading healthcare provider operating multiple hospitals across a major Indian region faced a critical challenge: patients needed a frictionless system to access medical consultation services. Their existing process required users to install apps or navigate complex websites, barriers that significantly reduced engagement, especially among the general population who prefer simple WhatsApp-based communication.
Managing 100–200+ daily inquiries manually was time-consuming and unsustainable. Scaling to 500–2000 daily inquiries would require a large team. Medical documents, voice notes, and text messages arrived scattered across channels, leading to manual data entry and inconsistent records.
A multilingual, multimodal WhatsApp automation system.
A2B built a multilingual, multimodal WhatsApp automation system powered by n8n, Gemini AI, and WhatsApp Business API, enabling patients to start consultations instantly through text, voice, or document uploads. The solution handles 100–200+ daily inquiries automatically, manages medical documents, collects structured patient information, and logs everything into a centralized database, without any manual staff involvement.
Phase 1: Multi-Modal Input Processing
Text & Voice Processing: Text messages are sent directly to the AI agent. Voice messages are downloaded from WhatsApp, transcribed to text using Gemini's audio API, and fed to the AI agent. Supports multilingual speech (English + Malayalam) enabling elderly users and patients with limited literacy to interact naturally.
Document Management: Prescriptions, lab reports, PDFs, images, and scans trigger a dedicated document pipeline for storage, tracking, and patient record updates.
Phase 2: Conversational AI Agent with Bilingual Support
Multilingual Support: Patients choose their language at the start, AI maintains it throughout the conversation, ensuring accessibility for all demographics in Kerala.
Structured Data Collection: The AI captures name, age, symptoms, hospital preference, type of service, appointment preferences, and document status.
Contextual Memory & Safety: The system retains the last 20 messages. The AI never gives medical advice, it only guides, collects data, clarifies queries, and prepares patient information for doctors.
Phase 3: Automated Document Management & Response Delivery
When patients share medical documents, files are downloaded from WhatsApp, uploaded to Google Drive, and links are generated for doctors. The patient database (Google Sheets) is updated, and the patient receives an acknowledgment message.
Based on conversation progress, professional WhatsApp messages are automatically sent in the patient's chosen language, providing next steps and offering hospital guidance. The automation eliminates manual data collection and transcription.
