Mira AI Shopping Assistant uses your product feed to recommend products, answer pricing questions, and check availability. This guide explains the feed formats Mira AI accepts, the tags it reads, and where to find the official export instructions for the most popular e-commerce platforms.
We've broken the guide into the following sections:
Before you start
Mira AI accepts standard XML product feeds — the same format most e-commerce platforms generate for Heuréka, Google Shopping, Zboží.cz and similar marketplaces. The richer your feed (price, availability, description, link, category), the smarter Mira's recommendations will be.
Tip: Once you have the feed URL, paste it as an XML product feed source in the Mira AI Sources tab — see Introduction to Mira AI for the full setup.
Note: Product card images shown in the chat are loaded from your product page (via link parser), not from the feed itself. Make sure your product pages have proper Open Graph or featured images.
Supported feed formats
Mira AI supports two standard product feed formats — Heuréka and Google Merchant. Both follow common XML conventions:
File extension is .xml or .rss.
The very first line must start with <?xml ...?> or <rss ...>.
One parent tag wraps all product items. The parent tag can have any name (e.g. <SHOP>) — only its existence matters.
Each product is wrapped in its own item tag (see the format-specific sections below).
A minimal valid skeleton looks like this:
<?xml version="1.0" encoding="UTF-8"?>
<SHOP>
<SHOPITEM> ... </SHOPITEM>
<SHOPITEM> ... </SHOPITEM>
</SHOP>
Important: If a mandatory tag is missing from a product, that whole product is skipped from Mira's index — visitors won't see it in recommendations. Make sure every product has all required tags filled in.
Heuréka feed structure
For the Heuréka format, Mira AI reads the following tags. See the official Heuréka XML feed documentation for the complete specification.
Category | Tags | Notes |
Item wrapper (mandatory) |
| One of these tags wraps each product. |
Description (mandatory) |
| Every tag must contain content. Empty tags count as missing. |
Voluntary |
|
|
Ignored |
| Mira AI skips these tags — you can leave them in the feed for other tools. |
A PARAM tag carries an extra product attribute as a name–value pair:
<PARAM>
<PARAM_NAME>Width</PARAM_NAME>
<VAL>100–109 cm</VAL>
</PARAM>
Google Merchant feed structure
For the Google Merchant format, Mira AI reads the following tags. See the official Google Merchant feed specification for the complete reference.
Category | Tags | Notes |
Item wrapper (mandatory) |
| One of these tags wraps each product. |
Description (mandatory) |
| For category, use |
Price format |
| Allowed format: |
Voluntary |
| In stock: |
Ignored |
| Mira AI skips these tags — you can leave them in the feed for other tools. |
Note: Numeric google_product_category IDs are not yet supported. Always use the text version of the category — for example "Apparel & Accessories > Clothing > Shirts & Tops" — or use product_type instead.
Export feed by platform
Find your e-commerce platform in the table below and follow the official export guide.
Platform | What it generates | Official guide |
Shoptet | Heuréka & Google Shopping XML from the admin panel | |
Shopsys | Product feeds via the platform with developer docs | |
Eshop-rychle | Feeds for comparison shopping engines | |
Upgates | XML data export with full product structure | |
Shoper | Google product feed directly from the platform | |
Shopware | Google Shopping — direct export or via add-ons | |
Shoprenter | Google Shopping & Heuréka feeds | |
PrestaShop | Google Shopping via add-ons | |
Magento | Google Shopping via extensions | |
WooCommerce | Google Product Feed via add-ons |
What's next?
Introduction to Mira AI Shopping Assistant — set up Mira AI and add your product feed as a source.
How to start training Mira AI — fine-tune AI answers with custom Q&A pairs.
Mira AI - stock availability — let visitors check whether a product is in stock.
Mira AI Shopping Assistant FAQ — common questions about Mira AI.