AI Product Recommendation for Bagisto: Complete User Guide

The Bagisto AI Product Recommendation extension keeps customers browsing. Showing the right items at the right time also keeps them buying.

Customers often leave a store when they cannot find what they want. This extension fixes that. It uses AI to learn about each product. Then it suggests similar ones on every product page.

Old related products need manual links. This extension uses AI embeddings. So it finds products with a close meaning.

For example, a customer who views a waterproof hiking boot sees outdoor gear and trail items. You never link them by hand.

Complete Features List

• AI picks – The extension finds similar products with AI.
• Two vector drivers – The admin keeps embeddings in Elasticsearch or MariaDB.
• Text fallback – The page still shows picks when AI data is missing.
• Native carousel – The extension uses the Bagisto product carousel, so it blends in.
• All visitors – Guests and logged-in customers both see the picks.
• Auto re-indexing – The index updates when a product changes, and a deleted product leaves it.
• All product types – Simple, virtual, configurable, grouped, bundle, downloadable, and booking products all work.
• Admin control – The admin sets everything in the panel, with no code.
• Any database – Bagisto can run on MySQL, MariaDB, or PostgreSQL, with or without a table prefix.
• 22 languages – Every label comes in all Bagisto languages.

Prerequisites

• Bagisto: v2.5.0
• PHP: 8.4 or later
• Database: MySQL, MariaDB, or PostgreSQL
• Vector driver: Elasticsearch 8.x, or MariaDB 11.7 or newer (with the VECTOR column type)
• MagicAI: installed, with at least one AI provider API key

Vector Driver Setup

First, install Elasticsearch and add its details to the .env file.

ELASTICSEARCH_HOST=your-url:9200
ELASTICSEARCH_USER=
ELASTICSEARCH_PASS=

For API key authentication, use these lines instead.

ELASTICSEARCH_HOST=your-url:9200
ELASTICSEARCH_API_KEY=your-api-key

For Elastic Cloud, use these lines.

ELASTICSEARCH_CLOUD_ID=your-cloud-id
ELASTICSEARCH_API_KEY=your-api-key

Next, set up the MariaDB driver only if you want it. It needs MariaDB 11.7 or newer.

This works when Bagisto runs on MySQL or PostgreSQL too. The vectors then live on a separate MariaDB server.

AI_RECOMMENDATION_MARIADB_CONNECTION=mariadb
AI_RECOMMENDATION_MARIADB_HOST=
AI_RECOMMENDATION_MARIADB_PORT=
AI_RECOMMENDATION_MARIADB_DATABASE=
AI_RECOMMENDATION_MARIADB_USERNAME=
AI_RECOMMENDATION_MARIADB_PASSWORD=

Without a MariaDB 11.7 connection, that driver stops with a clear message. So keep Elasticsearch in that case.

AI Product Recommendation Installation

You install it like any other Bagisto extension.

First, unzip the extension. Merge the packages/Webkul folder into your project root.

Next, open composer.json and add this line under ‘psr-4‘.

"Webkul\\AIRecommendation\\": "packages/Webkul/AIRecommendation/src"

Then refresh autoload.

composer dump-autoload

After that, open bootstrap/providers.php and add this line under Provider:

Webkul\AIRecommendation\Providers\AIRecommendationServiceProvider::class,

Also clear the caches.

php artisan optimize:clear

Finally, index your products. Use elasticsearch, mariadb, or both as the driver name.

php artisan ai-recommendation:index --driver=your-driver-name

Also add the fresh option to rebuild the index, for example after you change the AI model.

php artisan ai-recommendation:index --driver=your-driver-name --fresh

It has no migration. The MariaDB driver makes its own table on the first index run.

Admin Configuration

AI Provider Setup

This extension uses the Laravel AI SDK through the MagicAI package.

So it supports many AI providers, including OpenAI, Anthropic, Gemini, and Mistral.

Configure API Keys

First, go to Configuration>Magic AI>Providers.

Then enter the API key for your provider. Click Save Configuration.

AI Product Recommendation Configuration

Next, the admin goes to Configuration>AI Product Recommendation>General.

Status – When on, the carousel shows on every product page. When off, the store stays as it is.

Vector Driver – Pick where the embeddings live: Elasticsearch or MariaDB. Index your products into that driver first.

AI Model – Choose the AI model that makes the embeddings. If you leave it empty, the extension uses the MagicAI default.

Number of Recommendations – This sets how many products the carousel shows, from 1 to 20.

Minimum Similarity Score – The extension hides products that score below this value.

• A lower value, such as 0.2, shows more results but some may be less relevant.
• A higher value, such as 0.7, shows fewer results with more precise matching.
• The field takes one decimal place, so 0.7 works and 0.65 does not.

Section Title

In the Display section, the admin sets a heading for the carousel. If it is empty, the heading reads AI Recommended Products.

Customer Front End Panel

Product Page

A customer opens a product. A carousel of picks then shows below the product description.

It loads with the Related Products and Up-Sell Products carousels when the customer scrolls there.

Support

So, that was the user guide of the Bagisto AI Product Recommendation extension. For any doubts, write to us at [email protected].

You can also raise a ticket at our HelpDesk System.

Please see our Laravel Development Services and quality Bagisto Extensions.