In context
A scripted chatbot may return a fixed shipping answer. A tool-enabled shopping agent can search for a requested size, check available inventory and offer a permitted cart action. Retrieval-augmented generation is one common approach to providing store context, but not every agent retrieves on every turn or uses the same architecture.
A modern AI sales agent typically wires up four capabilities behind a single chat surface: semantic product search, structured tool use (add-to-cart, inventory check, order lookup, smart-offer), policy knowledge from your shipping/returns/FAQ content, and guardrails intended to limit unsupported claims and unsafe actions. These controls reduce risk; they do not make errors impossible.
The business goal isn't only deflecting support tickets; it's conversion. A well-tuned agent aims to turn more browsers into buyers, raise average order value through relevant cross-sells, and recover carts that would otherwise be abandoned. Measure that on your own store.
How Zubby uses this
Zubby is an AI sales agent for WooCommerce and custom or headless storefronts connected through the API on Pro or above. New Shopify connections are coming soon. We index your catalog and approved knowledge into store-scoped retrieval. Available product facts are ranked with your merchandising rules and can support product cards. Freshness depends on successful sync and indexing; custom-store facts depend on the latest API push. A product card does not mean every platform supports the same cart or order action.
Retrieved sources and tool calls are recorded in your conversation log, and the agent can hand off to a person when a shopper asks or when its confidence is low.