Conversational AI / Automation

WhatsApp Appointment Bot

A production-grade conversational AI agent that fully automates patient appointment scheduling via the Meta WhatsApp Business API.

WhatsApp Automation Dashboard

Role

AI Automation Eng.

Methodology

LLM Orchestration

Core Tech

n8n, Meta API

Impact

-70% Manual Effort

01

Context & Problem

Healthcare clinics face massive administrative bottlenecks managing patient appointments. Manual handling of bookings via phone calls or rudimentary text systems often results in missed appointments, accidental double-bookings, and frustrated patients.

Traditional "chatbot" solutions rely on rigid decision trees (press 1 for X, press 2 for Y), which provide poor user experiences and struggle with complex requests like rescheduling or checking specific availabilities.

The Objective

Engineer an intelligent, natural-language WhatsApp bot capable of handling end-to-end appointment workflows (booking, canceling, rescheduling) seamlessly, integrating directly with a live database.

02

Engineering Constraints

  • Multi-Turn State Management Unlike one-off queries, booking an appointment requires remembering context across multiple messages (e.g., handling slot conflicts, asking for patient details, then confirming).
  • Strict Latency Requirements To feel native to WhatsApp, the system needed an average response latency of under 2 seconds, requiring highly optimized LLM prompts and efficient async webhook processing.

03

Approach & Execution

I utilized n8n to orchestrate the complex webhook logic between the Meta WhatsApp Business API, the LLM intelligence layer, and the database.

🧠

LLM Dialogue Management

Implemented a system that parses natural language to extract precise intents (Book, Reschedule, Cancel) and entities (Dates, Times, Patient Names) flawlessly.

📊

Real-Time Database Sync

Integrated Google Sheets as a lightweight, real-time CRM to manage over 500 patient records and dynamically query available slots to prevent double bookings.

To ensure reliability, I engineered automated error-handling fallbacks within the n8n workflows, guaranteeing that edge cases or LLM timeouts would gracefully guide the patient rather than dropping the conversation.

04

Impact & Value

The deployment of this system transformed chaotic manual booking processes into a streamlined, 24/7 autonomous operation.

70%

Manual Effort Reduction

Drastically cut down the time administrative staff spent on the phone scheduling and modifying patient appointments.

99%

Workflow Completion

Achieved near-perfect successful workflow completion rates thanks to strict async webhook management and edge-case fallbacks.

Ready to see more of my work?