Marlon Mejía
GitHub ↗Back to projects

AI workflow automation · Functional demo

WhatsApp
AI Chatbot

A real-time conversational automation that connects WhatsApp, n8n, Google Gemini, and session memory in one validated, end-to-end workflow.

Watch the demo Explore GitHub
  • n8n
  • Google Gemini
  • WhatsApp Cloud API
  • Docker
  • Cloudflare Tunnel
LIVE WORKFLOWV1 · DEMONSTRATION
Animated demonstration of a WhatsApp conversation activating the n8n workflow and returning an AI-generated response
Message → validation → AI Agent → replyFunctional
Real-timeMessage processing
Session-basedConversation memory
ValidatedWebhook events
Database-backedVersion 2 roadmap

01 — Project purpose

Turning a message into an intelligent automated action.

This functional demonstration shows how conversational AI can become part of a reliable business workflow instead of operating as an isolated model.

  • Receive and process messages through the official WhatsApp Cloud API.
  • Validate events before invoking the AI workflow.
  • Maintain separate context for each WhatsApp sender.
  • Generate and deliver contextual responses automatically.

02 — Architecture

A complete loop from webhook to response.

n8n coordinates the integrations while the AI Agent combines Gemini with conversation memory.

  1. 01WhatsAppIncoming user message
  2. 02Cloud APIWebhook event delivery
  3. 03n8nValidation and orchestration
  4. 04AI AgentGemini + session memory
  5. 05ResponseAutomated WhatsApp reply
Supporting servicesGoogle GeminiSession MemoryCloudflare Tunnel

03 — Workflow design

Simple on the surface. Deliberate underneath.

The visual workflow keeps every responsibility visible: trigger, validation, intelligence, memory, and delivery.

n8n workflow with a WhatsApp trigger, event validation, Google Gemini AI Agent, simple memory, and send-message node
Sanitized public workflowCredentials and sensitive values excluded

04 — What it demonstrates

01

Event validation

Incoming webhook events are checked before the AI flow runs, preventing unsupported events from triggering unnecessary executions.

02

Contextual responses

The AI Agent combines Google Gemini with sender-based session memory to keep each conversation coherent across messages.

03

Automated delivery

The generated answer returns through the WhatsApp Cloud API, completing the flow without manual intervention.

04

Secure connectivity

A Dockerized n8n environment and Cloudflare Tunnel provide the public HTTPS endpoint required for webhook communication.

05 — Functional demo

Watch the workflow respond in real time.

The video shows the message, the n8n execution, conversational context, and the automated response delivery.

DEMO RECORDING02:02

06 — Version 2

From short-term context to persistent intelligence.

The current version proves the full integration. The next version adds a persistent database layer so conversation history can be stored, retrieved, and used over much longer periods.

In developmentV2
  • 01Persistent conversation history
  • 02Database-backed users and sessions
  • 03Long-term information retrieval
  • 04Improved context management
  • 05Scalable foundation for future AI capabilities
Follow the project on GitHub ↗