Adding an AI Sales Assistant to a Shopify Store

Shopify merchants have no shortage of chat-widget apps to choose from, and most of them share the same limitation: they're a chat window bolted onto the storefront, disconnected from real product data, answering from whatever the operator manually configured. A widget that can't check live stock or pull a real price isn't a sales tool — it's a slightly smarter contact form.

What a real integration needs to do differently

The bar for a genuinely useful storefront assistant is the same bar that applies to WhatsApp sales conversations generally: every product question needs to be answered from the actual catalog, not a canned script. A visitor asking "is this in stock in medium?" needs the real answer, sourced live. If the underlying assistant is the same one handling a merchant's WhatsApp conversations, the storefront widget isn't a separate product with its own logic to maintain — it's the same deterministic pipeline, catalog lookups and all, running on a different channel.

The install itself shouldn't be an engineering project

A common failure point for Shopify integrations is the install process: app store review queues, OAuth flows, custom app development. For a merchant, that's days of friction before the tool does anything useful. In practice, adding Sellatrix's widget to a Shopify storefront is two `<script>` tags added directly to the theme — one setting the widget's public key, one loading the chat bundle — pasted in once by anyone with theme-editing access. No app-store review process, no custom app development. The widget then runs in an isolated Shadow DOM, so it can't clash with the store's own styles in either direction.

Security and scope, not just convenience

An embedded widget is public-facing by definition, which makes a few things non-negotiable: the widget's key can be restricted to only accept requests from the store's own domain (an origin allowlist enforced server-side, not just client-side), and it needs rate limiting per visitor to prevent abuse — both without requiring the merchant to think about either. Visitor continuity matters too: a shopper who reloads the page or comes back later resumes the same conversation rather than starting over, via a persisted session identifier rather than one that resets on every page load.

Why this is worth doing on Shopify specifically

Shopify stores already have structured, queryable product data — variants, pricing, inventory — sitting in a format built for exactly this kind of lookup. That's what makes a catalog-grounded assistant straightforward to stand up on Shopify in a way it isn't on a hand-built site with no structured product layer: the hard part, having reliable data to answer from, is already solved by the platform. The integration work is mostly about connecting to it cleanly — via a guided authorization flow, not a manual data migration — not building a product database from scratch.