WhatsApp AI Sales Agent with Product Catalog
N8N
Project's Category
Intro
This one runs product sales through WhatsApp instead of a website. A customer can text a question, send a voice note, or drop in a photo of something they're eyeballing, and the agent answers back with real product details pulled from a Supabase knowledge base.
Problems
The client was getting the same product questions over and over on WhatsApp, price, availability, specs, and someone had to be sitting there answering every single one. Voice notes and photos made it worse, since those can't just be searched like plain text.
Solutions
We connected n8n to a Supabase catalog holding SKUs, prices, descriptions, and image links, then built an agent that reads incoming WhatsApp messages in whatever format they arrive. Voice notes get transcribed, images get run through recognition to identify the product, and everything gets matched against the catalog and FAQs before the agent replies. It was originally built for a furniture business, but the setup swaps to almost any catalog style business without much rework.
Final Thoughts
The fallback response node is doing quiet, important work here. If nothing matches well enough, the customer gets a real message instead of a wrong answer or dead silence. Small thing, but it's the difference between a slightly annoying bot and a broken one.
WhatsApp AI Sales Agent with Product Catalog
N8N
Project's Category
Intro
This one runs product sales through WhatsApp instead of a website. A customer can text a question, send a voice note, or drop in a photo of something they're eyeballing, and the agent answers back with real product details pulled from a Supabase knowledge base.
Problems
The client was getting the same product questions over and over on WhatsApp, price, availability, specs, and someone had to be sitting there answering every single one. Voice notes and photos made it worse, since those can't just be searched like plain text.
Solutions
We connected n8n to a Supabase catalog holding SKUs, prices, descriptions, and image links, then built an agent that reads incoming WhatsApp messages in whatever format they arrive. Voice notes get transcribed, images get run through recognition to identify the product, and everything gets matched against the catalog and FAQs before the agent replies. It was originally built for a furniture business, but the setup swaps to almost any catalog style business without much rework.
Final Thoughts
The fallback response node is doing quiet, important work here. If nothing matches well enough, the customer gets a real message instead of a wrong answer or dead silence. Small thing, but it's the difference between a slightly annoying bot and a broken one.
Building the automation systems that handles business processes.
Clear Communication
Reliable
Professional
Reliable Quality
Building the automation systems that handles business processes.
Clear Communication
Reliable
Professional
Reliable Quality
