
Overview: A voice agent that recommends products in real time by understanding the entire catalog.
š Problem: Shoppers ask for recommendations in natural language ("something for dry skin under $30"), but standard search can't reason over a catalog or answer conversationally ā so customers bounce.
š” Solution: A Retell voice agent backed by retrieval-augmented generation (RAG): the whole product catalog is embedded into a vector store, and an AI agent retrieves the most relevant items and answers conversationally over a webhook.
Workflow Architecture:
⢠Catalog embedding ā product data embedded with OpenAI, loaded into a vector store.
⢠Voice layer (Retell) ā natural, real-time voice conversation.
⢠Recommendation agent (RAG) ā retrieves the most relevant items for the shopper's request.
⢠Storefront wiring ā connected to a live Shopify catalog over a webhook.

Workflow Architecture

Workflow Image
š Tools: Retell (Voice AI) Ā· OpenAI embeddings Ā· Vector store (RAG) Ā· n8n AI Agent Ā· Shopify Ā· Webhooks
āļø Results: Conversational product discovery in real time Ā· recommendations grounded in the real catalog (no hallucinated products) Ā· works hands-free over voice.
š Impact: Turns a catalog into a knowledgeable assistant that guides shoppers to the right product ā the kind of help that lifts conversion and average order value.
0 likes Ā· 5 views Ā· Jun 22, 2026
Developed a voice agent for product recommendations using AI and Shopify.