vviinn Documentation
Infrastructure for Agentic Commerce
Section titled “Infrastructure for Agentic Commerce”vviinn is a suite of AI-powered search and discovery services for online shops. You provide your product data; vviinn’s AI models (the Nets) turn it into vector embeddings — a representation of what each product actually is, in meaning and appearance, not just its raw text and images. From there, each service matches those embeddings against a different kind of shopper intent: a typed query, an uploaded photo, “more like this,” or “goes well with this.” You configure how each one behaves for your catalog through Panel, vviinn’s merchant-facing admin interface.
This documentation explains what each service does and how to set it up in Panel.
What you’ll need to do
Section titled “What you’ll need to do”Getting vviinn running in your shop comes down to three things:
- Provide a product feed — a URL to your product data that vviinn imports and keeps in sync.
- Configure services in Panel — turn on and tune whichever services below you want to use.
- Embed the widget — a small JavaScript snippet that adds the search bar, recommendations, and other shopper-facing UI to your storefront. Building your own front-end instead? Call the Platform API directly — see How It Fits Together for details. Building a custom AI agent instead? Integrate vviinn search via the MCP tools.
See How It Fits Together for how these pieces connect to each other.
Services
Section titled “Services”Each service page ends with a Configuring in Panel section that walks through the exact screen and settings.
-
Keyword and semantic product search for your search bar, with query understanding and auto-suggest.
-
Shoppers search with a photo — matched by visual similarity, text in the image, or a barcode.
-
“More like this” recommendations for a product page.
-
“Goes well with this” recommendations for complementary products.
-
🏷️ Product Data
AI-predicted product attributes, with a review workflow to confirm or correct them.
-
📈 Boosting
Re-ranks results by your own business signals — margin, preferred brands, price direction — without hiding any results.
How it fits together
Section titled “How it fits together”Shoppers interact with the vviinn Widget (documentation) embedded in your shop; you configure everything through Panel. See How It Fits Together for the full picture.
How-to guides
Section titled “How-to guides”Step-by-step guides for common setup and integration tasks:
| Guide | What it covers |
|---|---|
| Panel: Prepare a Product Feed | The delimited-text file format your catalog needs, and the questions that come up preparing one — CSV vs. TSV, multiple feeds, custom columns |
| Widgets: Search & Results Page Redirects | Send shoppers from your search bar to a dedicated results page, with shareable URLs for both text and image searches |
Developer tools
Section titled “Developer tools”If you’re integrating an AI agent or a custom application against vviinn directly (rather than through the widget), see Developer Tools (MCP) for the Model Context Protocol tools that expose search, recommendations, and catalog data programmatically.
Other documentation
Section titled “Other documentation”This site covers services, Panel configuration, and the widgets. A couple of other vviinn surfaces have their own documentation:
- Widget playground — a Storybook for trying every widget and its attributes interactively, alongside the widget changelog.
- Platform API — the full endpoint reference (OpenAPI spec) for
api.vviinn.comis in the API Reference, useful if you’re calling it directly instead of using the Widget. - Panel itself — Panel has a built-in AI Assistant in its navigation for questions about your specific configuration, in addition to this documentation.