Trader Information
Back to topOverview
Back to topThe MageDelight AI Product Recommendations extension uses Chroma as its retrieval engine: catalog data is converted into embeddings, stored in Chroma, and queried in real time whenever a recommendation block renders. A growing fashion retailer, for example, can enable browsing-based recommendations on category pages and wishlist-based recommendations in the account area, both staying current as behavior changes with no manual list-building. Shoppers see blocks styled to match the store's existing theme; in the admin, each block's accuracy threshold, product limit, and visibility can be adjusted independently, with debug mode available to inspect match percentages.
Account & Pricing
Chroma does not require a paid account in the way a typical SaaS API does — it is free, open-source software. The extension does not create or configure a Chroma instance automatically during installation; the merchant or their technical team sets one up separately, either self-hosted at no cost using Chroma's getting-started guide, or through the managed Chroma Cloud offering, which is usage-based and billed directly by Chroma, separate from the extension's purchase price. MageDelight support covers extension-side configuration once a Chroma instance is reachable.
Features:
- Unified recommendation engine across home, category, product, cart, and wishlist placements
- Recommendations based on wishlist activity
- Recommendations based on past purchases and replenishment cycles
- Recommendations based on live browsing and search behavior
- Contextual recommendations on the current product page
- Admin control over recommendation accuracy and product limit per block
- Debug mode showing match percentages behind each suggestion
- Automatic exclusion of out-of-stock and already-owned items
- Diversity rules and exploration boosts to avoid repetitive suggestions
Compatibility spans Open Source, Adobe Commerce, and Adobe Commerce Cloud, versions 2.4.0.x through 2.4.9.x, with native rendering on Hyvä-compatible themes.
Business-Level Overview
For an eCommerce manager, the gap between a static "related products" block and a genuinely personalized storefront shows up directly in conversion and average order value. This extension closes that gap with recommendations that adjust to each shopper's own browsing, wishlist, and purchase signals.
Operationally, the model removes a recurring merchandising task: no one has to hand-curate cross-sell lists or refresh them as inventory turns over. Guardrails and diversity rules stay in the merchandising team's control, but day-to-day ranking runs on its own.
Revenue impact follows from matching intent more precisely: typical conversion rate lifts in the 5-15% range are reported for personalized recommendations, alongside higher average order value and faster discovery that reduces bounce.
For a founder or CTO evaluating fit, the underlying vector-search approach scales with catalog size, and inventory-aware filtering keeps suggestions trustworthy by never promoting something a shopper can't buy.
3rd Party Integrations
API Provider: Chroma
What it enables: Chroma is the open-source vector database that stores product embeddings and performs the similarity search behind every recommendation. It must be installed and configured, self-hosted or via Chroma Cloud, for the extension to generate matches.
Authentication method: Connection URL to a self-hosted Chroma instance, or an API key when using the hosted Chroma Cloud deployment.
Storefront User Benefits
- Faster product discovery matched to real browsing intent
- Wishlist and past-purchase items resurfaced automatically
- Relevant, in-stock alternatives instead of dead ends on sold-out items
- A more personal shopping experience from homepage through checkout
- Fewer irrelevant suggestions cluttering product and cart pages
Admin Benefits
- Fine-tune recommendation accuracy and per-block product limits from the admin
- Enable debug mode to view match percentages behind any recommendation
- Set diversity rules and exploration boosts to keep suggestions varied
- Automatically exclude out-of-stock or already-owned items with no manual list upkeep
- Control which recommendation types appear on which page, storefront-wide
Full Feature List
- Unified AI-driven product recommendation engine
- Recommendations based on wishlist activity, past purchases, and browsing
- Contextual recommendations on the current product page
- Admin-configurable accuracy threshold and product limit
- Debug mode with visible match percentages
- Inventory-aware filtering of out-of-stock items
- Suppression of already-owned or recently returned items
- Restock-aware resurfacing of previously suppressed items
- Diversity rules and exploration boosts for merchandising control
- Hyvä-compatible storefront rendering
Business Problem vs Solution
Static blocks ignore individual intent Problem: One-size-fits-all related-product sections show the same items to every visitor. Solution: Live affinity profiles from browsing, wishlist, and purchase signals rank products by individual likelihood to buy.
Manual curation can't keep pace with change Problem: Hand-built cross-sell lists require constant upkeep and go stale as stock and demand shift. Solution: Recommendations update automatically with the catalog and shopper behavior, with no manual list-building.
Sold-out or already-owned suggestions erode trust Problem: Generic logic keeps surfacing unavailable items or products a shopper already purchased. Solution: Inventory-aware filtering excludes out-of-stock SKUs and suppresses already-owned or returned items.
No visibility into why a recommendation was made Problem: Black-box suggestion logic leaves teams unable to audit the results. Solution: Debug mode exposes match percentages behind every suggestion for verification.
Business Benefits (Measurable Outcomes)
- 5-15% typical lift in conversion rate from personalized, intent-matched recommendations
- Higher average order value through cross-sell and complementary-item suggestions
- Faster product discovery that reduces bounce and time-to-cart
- Inventory-aware logic that avoids promoting out-of-stock SKUs
- Less manual merchandising effort, freeing teams to focus on strategy
- Improved repeat-purchase rate through replenishment-aware suggestions
Sample Use Case
A multi-store fashion retailer (D2C) relied on manually curated "you may also like" blocks, updated in spreadsheets and pushed live weekly, a process that lagged behind real inventory and seasonal shifts.
After installing the MageDelight AI Product Recommendations extension and connecting it to a self-hosted Chroma instance, the team enabled browsing-based recommendations on category pages, wishlist-based recommendations in the account area, and contextual recommendations on product pages, all configured through the admin.
Within weeks, average order value increased through stronger cross-sell placements, the weekly manual task was eliminated, and debug mode let the team verify match accuracy before rolling out to every storefront in the group.
FAQs
Q1: What is the AI Product Recommendations extension and how does it work? It builds a live affinity profile from browsing, wishlist, and purchase activity, then queries a Chroma vector database to rank relevant products in real time.
Q2: Which store versions are supported? Open Source, Adobe Commerce, and Adobe Commerce Cloud, versions 2.4.0.x through 2.4.9.x, with native rendering on Hyvä-compatible themes.
Q3: Does installing this extension require custom development? No. Blocks, accuracy, and product limits are configured from the admin; setting up a separate Chroma instance is the only external step.
Q4: Can I control where recommendations are displayed? Yes. Each type can be enabled or disabled per placement, with product limits set independently per block.
Q5: Does this extension require ongoing service fees or API keys? Only if the connected Chroma instance requires one. Self-hosted Chroma is free; Chroma Cloud is usage-based and billed separately.
Q6: Will this extension slow down my storefront? No. Embedding lookups run against the connected Chroma instance; the storefront only renders the returned product list.
Q7: How do AI recommendations outperform manually curated lists? They rank per shopper instead of one fixed list, update automatically, and exclude out-of-stock or already-owned items.
Technical Specifications
Back to topSeller profile
Seller contact
Current Version
1.0.3
Adobe Commerce platform compatibility
Adobe Commerce (cloud): 2.4 (current)
Adobe Commerce (on-prem): 2.4 (current)
Magento Open Source: 2.4 (current)
Type
Stable Build
Updated
22 October, 2026
Categories
Extensions, Sales
Documentation
License Type
GNU General Public License 3.0 (GPL-3.0)
Policy
Quality Report
Back to topAll tests were conducted on the latest versions of Adobe Commerce that existed for the compatible release lines at the moment of the extension submission. Latest versions of all other software were used, as applicable.
Release Notes
Back to top1.0.3:
- Compatible with Adobe Commerce (cloud) : 2.4
- Compatible with Adobe Commerce (on-prem) : 2.4
- Compatible with Magento Open Source : 2.4
- Stability: Stable Build
-
Description:
Stable Release on September 01, 2026
Support
Back to topThe best place to start if you need help with a specific extension is to contact the developer. All Adobe Commerce developers have both a contact email and a support email listed.
Contact Vendor