Product recommendations are one of the highest-ROI features in e-commerce. Amazon attributes up to 35% of its revenue to its recommendation engine. Yet most Magento 2 stores still rely on rule-based widgets "frequently bought together" lists that are manually curated, quickly outdated, and entirely static. They show the same products to every visitor, regardless of what they're actually browsing.
Mavenbird's AI Product Recommendation Extension changes that. It uses AI embedding technology to analyse product names, SKUs, attributes, descriptions, and metadata and surfaces genuinely similar products in real time. Every recommendation is dynamically generated, configurable in accuracy, and works for both guest shoppers and registered customers without any manual curation.
AI-powered product recommendations for Magento 2 using embedding-based matching grid & slider views, configurable accuracy, GPT-5 support, guest & registered user personalisation.
What is the AI Product Recommendation Extension?
The Mavenbird AI Product Recommendation Extension is a Magento 2 module that brings genuine AI-driven personalisation to your product pages, category pages, and any other location in your store without requiring manual curation or rule setup.
Unlike traditional recommendation widgets that rely on manually defined rules ("show products from the same category") or co-purchase frequency data that takes months to accumulate, this extension uses AI embedding technology via ChromaDB to compute product similarity mathematically. Products are converted into high-dimensional vectors based on their attributes and metadata and the AI finds the closest matches in that vector space, returning the most relevant suggestions for whatever a shopper is currently viewing.
The Business Case What AI Recommendations Deliver
How the AI Recommendation Engine Works
When a product is saved or the extension is installed, the AI indexes each product's name, SKU, attributes, description, and metadata. This data is passed to the AI model and converted into a high-dimensional vector embedding that mathematically represents the product.
Product embeddings are stored in ChromaDB a purpose-built vector database optimised for high-speed similarity search. ChromaDB enables the extension to query millions of product vectors in milliseconds, making real-time recommendations possible at any catalogue scale.
When a shopper views a product page, the extension queries ChromaDB with the current product's embedding. ChromaDB returns the nearest-neighbour products those with the highest vector similarity within the configured accuracy threshold (distance value).
The matching products are rendered in your configured display format grid or slider with the number of products and layout controlled by admin settings. Both guest and registered users receive recommendations instantly on every eligible page.
As new products are added or product data is updated, embeddings are regenerated automatically. The recommendation engine stays current with your catalogue at all times no manual re-curation required.
Two Display Modes Grid & Slider
The extension offers two flexible frontend display modes, both configurable from the Magento admin and placeable on any page of your store:
- Multiple recommended products displayed in configurable column rows
- Shows product image, name, price, and rating at a glance
- Adapts automatically for desktop, tablet, and mobile breakpoints
- Ideal for product pages and category landing pages
- Place anywhere on your storefront with flexible widget positioning
- Interactive horizontal slider for smooth product browsing
- Auto-cycle through recommendations without user input
- Navigation arrows and dots for easy manual control
- Ideal for homepage carousels, cart pages, and high-traffic landing pages
- Keeps users engaged and surfacing more of your catalogue
Key Features
Recommendations generated by analysing product names, SKUs, attributes, descriptions, and metadata using AI embedding models. Relevance is mathematically computed not manually set ensuring genuinely related suggestions for every product.
Admins control the recommendation precision via a distance value setting. A lower value returns highly similar products; a higher value broadens the scope. Fine-tune to match your catalogue's density and the user experience you want to deliver.
Every shopper whether they have an account or not receives personalised AI recommendations. No login required for personalisation, meaning your recommendations work from the very first visit.
Two built-in frontend layouts a product grid for discovery-mode pages and an interactive slider for high-traffic pages and homepages. Both are responsive and configurable from the admin without code changes.
Configure which AI recommendation sliders appear on which websites and store views. Each slider can be scoped to specific store views perfect for multi-store operations serving different markets or languages.
Full GraphQL and REST API support for headless and PWA storefronts. Slider lists respect store visibility configuration through both API interfaces, ensuring consistent personalisation across all frontend architectures.
Understanding the Distance Value Your Accuracy Control
One of the most powerful and most misunderstood features of the extension is the distance value. This setting directly controls how strictly the AI matches products to the one currently being viewed. Here's how to think about it:
Only very closely related products are shown. Best for stores with large catalogues where loose recommendations would confuse shoppers.
A good middle ground relevant without being too narrow. Suitable for most Magento stores as a starting configuration.
Wider product discovery useful for stores with smaller catalogues where strict matching would return too few results.
How to Install & Set Up the Extension
Purchase the AI Product Recommendation Extension from store.mavenbird.com. Your licence key and extension package will be delivered via email. Community edition is $199; Enterprise/Enterprise Cloud is an additional $199.
Unzip the extension package and upload the module folder to your Magento installation under the app/code/Mavenbird/ directory.
Navigate to Stores → Configuration → Mavenbird → AI Product Recommendations. Set your ChromaDB server endpoint, configure the number of recommendations to display, and set your initial distance value. Save configuration.
Navigate to Mavenbird → AI Product Recommendations → Manage Sliders. Create a new slider, select Grid or Slider display mode, set which store views to target, and configure your layout preferences (products per row, navigation controls, etc.).
Add the AI recommendation block to your product pages, category pages, or homepage using the widget code, CMS block, or Page Builder widget. Flush cache and your AI recommendations are live.
AI Recommendations vs. Rule-Based Recommendations
| Capability | Rule-Based (Manual) | Mavenbird AI Extension |
|---|---|---|
| Recommendation logic | Fixed rules (same category, same brand) | ✓ AI embedding similarity contextually aware |
| Setup time | Hours/days of manual rule configuration | ✓ Automated works on install |
| Maintenance | Constant manual updates as catalogue changes | ✓ Self-updating re-indexes automatically |
| New products | Must be manually added to rules | ✓ Indexed automatically on save |
| Cross-category recommendations | ✗ Not possible | ✓ Yes AI finds related products across categories |
| Guest user personalisation | ~ Limited | ✓ Full recommendations for all visitors |
| Accuracy tuning | ✗ Not available | ✓ Distance value fine-tune precision |
| Headless / GraphQL support | ~ Depends on implementation | ✓ Full GraphQL & REST API support |
| Store view targeting | ~ Manual per-view setup | ✓ Built-in store view scoping per slider |
v2.5.0 (March 2026)
Who Benefits Most?
Fashion & Apparel Recommend complementary items across categories (shoes with dresses, belts with trousers) that a rule-based system could never surface. Increase units-per-transaction across every session.
Electronics Suggest compatible accessories, upgrades, and peripherals based on the technical similarity between products cables, cases, adapters, and warranties surfaced intelligently, not manually.
Home & Furniture Recommend lifestyle-matched products across a broad range of categories. A shopper viewing a dining table sees matching chairs, centrepieces, and placemats not just other tables.
Health & Beauty Complement skincare product views with routine-compatible serums, toners, and moisturisers based on ingredient and benefit similarity driving meaningful basket building.
B2B Distributors Surface compatible parts, consumables, and accessories for technical products relationships a human would need deep product knowledge to define, but that the AI maps automatically from product data.
Frequently Asked Questions
How does the Magento 2 AI Product Recommendation extension work?
The extension uses AI to convert product data names, SKUs, attributes, descriptions, and metadata into high-dimensional vector embeddings stored in ChromaDB. When a shopper views a product, the extension queries ChromaDB to find the most similar product embeddings and renders those as recommendations. Admins configure the count of products shown and the distance value (accuracy threshold).
What is the distance value and how should I set it?
The distance value defines how closely matched recommended products must be to the viewed product. A value of 1 returns only very similar products (high precision). A value of 10+ returns a wider range (broad discovery). Start around 3–5 for most stores and adjust based on recommendation click-through performance. Lower the value if recommendations feel too loose; increase it if the block shows too few products.
Does the extension work for guest shoppers?
Yes. AI recommendations are delivered to both guest users (not logged in) and registered customers. Personalisation is based on the product being viewed not on the user's account history so every visitor receives relevant recommendations from their very first page view.
Is it compatible with headless Magento / PWA storefronts?
Yes. The extension includes full GraphQL and REST API support. Recommendation slider lists respect store visibility configuration through both API interfaces, making it compatible with headless commerce architectures, Vue Storefront, PWA Studio, and custom React/Next.js frontends.
Is the extension compatible with Hyvä Theme?
Yes. The extension is officially Hyvä Ready and has been tested on Hyvä-based storefronts. It is also compatible with standard Luma themes, Magento 2 CE, EE, Adobe Commerce, and PHP 8.4 on versions 2.3.x and 2.4.x.
Can I show different recommendation sliders on different store views?
Yes. Version 2.0.0 introduced store view targeting each AI recommendation slider can be configured to display only on specific websites and store views. This is essential for multi-store operations serving different markets, languages, or product catalogues from a single Magento installation.
What is included with the $199 purchase?
The $199 price covers Magento Community Edition. Enterprise and Enterprise Cloud editions are available for an additional $199 each. Every purchase includes free installation support, one year of free technical support, and one year of free updates including compatibility patches for new Magento versions released during that period. A 30-day money-back guarantee applies.
Ready to put AI recommendations on your Magento 2 store?
The Mavenbird AI Product Recommendation Extension is $199 one purchase, free installation, 1 year of support & updates, 30-day money-back guarantee. Try the live demo before you buy.
