E-commerce WhatsApp Assistant
First-line WhatsApp support with dynamic catalog, shipping quotes and human handoff.
Role
System design and implementation
Problem
The brand needed a scalable first line on WhatsApp for repetitive product and shipping questions, without hard-coding business rules inside automation flows.
What I developed
A WhatsApp assistant that reads from a dynamic catalog, quotes shipping through an API, applies editable business rules and escalates sensitive cases to a human operator.
Architecture
WhatsApp channel → n8n orchestration → catalog layer (Google Sheets) → LLM response layer → Redis for session context → shipping API → human handoff path.
What it handled
Product Q&A, shipping quotes, catalog lookups, session context and escalation to human support when rules require it.
Key decisions
- Keep business rules editable outside the workflow
- Use Redis for lightweight session state
- Define explicit human handoff triggers
Evidence available
Architecture diagram · sanitized conversation examples · production notes
Limitations
Public documentation uses sanitized examples only. No client names, real conversations or operational metrics are exposed.