AI Product Support is a Magento 2 module that adds a convenient AI chat to the store.
It can work in two places:
Its purpose is to quickly answer questions about products, modules, documentation, and supporting content stored in the AI knowledge base prepared for the store.
This is an important distinction: this module does not organize store content from scratch. It makes existing knowledge available through a convenient chat.
In everyday work, the module helps reduce the time spent searching for information. Instead of browsing descriptions, instructions, FAQs, and notes in different places, the user can simply ask a question in the chat.
Example use cases:
The module was designed to work in different types of Magento 2 stores, for example:
This is not a solution for only one specific project. It is a module that can be adapted to many stores, provided they have a prepared knowledge base for AI.
For this kind of chat to work well, the store first needs organized content.
In practice, this means materials such as:
On the Kowal AI Product Feed module page, this is described very accurately: its role is to turn store product data into an organized, up-to-date knowledge base ready to be used by AI systems. This is the stage that prepares the foundation for later AI features in the store. Source: Kowal AI Product Feed for OpenAI Vector Store
The simplest way to understand it is this:
WhatsApp
+48608012047AI Product Support uses that knowledge in a conversation with the user.As a result, the chat does not operate separately from the store. It uses materials prepared specifically for a given offer.
A permanent AI tab appears in the Magento panel at the right edge of the screen. It is visible while working in different sections of the panel and opens a slide-out chat panel.
This gives the user access to AI assistance from almost anywhere in the panel:
The module can also be enabled on the store frontend. The customer then sees a simple AI tab and can ask a question without leaving the page.
This solution is especially useful where customers often ask about:
The module does not work like a regular general AI chatbot. It was prepared to answer primarily based on the knowledge provided by the store.
This means:
This is the biggest difference between a regular AI chat and a solution implemented sensibly in a store. First, you prepare the knowledge, and only then make it available to users as a chat.
Answers are presented in a convenient format:
If the question concerns a specific product, the module can additionally show a section with product cards, for example:
The module has its own configuration, so it can be managed independently from other extensions.
The administrator can set, among other things:
If the chat runs on the store frontend, the module helps limit unnecessary load and accidental abuse.
It includes, among other things:
This helps better manage costs and operational stability.
An employee opens the AI tab, enters a question, and receives an answer related to the store offer. They can ask, for example, about:
The customer can use the chat as a simple information assistant. Instead of clicking through many pages, they receive an answer faster and can move on to the right product.
The module requires:
This document explains how to launch the AI Product Support module in a Magento 2 store and how to prepare it for use.
The document is written for a store that wants to implement a ready-made module and start using AI chat without going into the technical details of how it works on the code side.
It is best to treat this module as a user-facing layer. First, the store prepares the knowledge, and only then makes it available to employees or customers as a chat.
Before implementation, you need:
The most important practical requirement is simple: the module will provide good answers only when the store has prepared content that AI can use.
If the store already uses the Kowal AI Product Feed module, it can serve as the tool for preparing and organizing content for AI. On that module page, this is described as building an organized and up-to-date knowledge base for AI systems. AI Product Support is the natural next step, meaning the use of that knowledge in a conversation with the user. Source: Kowal AI Product Feed for OpenAI Vector Store
The module is installed via composer.
Example installation process:
composer config repositories.ai.product.support vcs https://github.com/kowalco/ai-product-supportcomposer config --global --auth github-oauth.github.com composer require kowal/module-ai-product-supportbin/magento module:enable Kowal_AiProductSupportbin/magento setup:upgradebin/magento cache:clean If the store runs in production mode, also execute after installation:
bin/magento setup:di:compilebin/magento setup:static-content:deploy -fbin/magento cache:cleanAfter installation, you can find the module settings here:
Stores > Configuration > Kowal AI > AI Product Support
In the General section, enable:
Enable ChatAfter saving the setting, the AI tab will appear in the admin panel.
If you want store customers to be able to use the chat as well, enable:
Enable Frontend ChatIf the module is intended only for the store team, leave the frontend disabled.
The field:
Maximum Question Lengthdefines the maximum length of a message the user can send. This helps keep things organized and limit overly long, unreadable requests.
If the store operates in multiple language versions or has several store views, you can specify:
Default Store ViewThis makes work in the panel easier and helps start from the correct context.
The field:
Allow Store Switcherdetermines whether the panel user can change the store view directly in the chat popup.
This is useful when one team supports several store versions.
The field:
Log Chat Requestsis worth enabling during configuration and testing. This makes it easier to check whether the module works correctly. After the production rollout, you can decide whether logging should remain active.
In the OpenAI section, configure the basic elements required for the module to work.
OpenAI API KeyThis is the access key to the AI service. Without it, the module will not retrieve the list of models, read the knowledge source, or send a question.
Response ModelThis is the model responsible for building answers.
The model list is loaded from the API. If no options appear after saving the key, you can use the button to refresh the model list.
Vector StoreThis is the selected knowledge source for the module.
The simplest way to understand it is this:
If the store already has an organized knowledge base prepared for AI, this is where you select that source.
The list of knowledge sources can also be refreshed from the configuration level.
Maximum File Search ResultsThis setting defines how many supporting materials the module takes into account when preparing one answer.
In practice, it affects:
A good starting setting is a medium value, for example 6.
If the chat runs on the store side, it is worth setting security limits right away.
In the Frontend Security section, you will find:
Requests Per MinuteRequests Per HourMinimum Submit DelayThese settings help limit:
In the Prompting section, you can fill in:
System Prompt TemplateThis field is not required. In most implementations, you can leave it empty and use the module default setting.
Overriding it makes sense only when the store wants to introduce its own answer style or additional communication rules.
After the module is enabled, the user sees an AI tab at the right edge of the screen. Clicking it opens the chat panel.
In the panel, you can:
If the chat has been enabled, a similar tab appears on the store side. The customer can ask a question without leaving the product page or listing.
The most practical model looks like this:
This approach gives a better result than launching the chat alone without prepared data.
Enable Chat = YesEnable Frontend Chat = depending on the projectMaximum Question Length = 1000Allow Store Switcher = Yes for multiple store viewsLog Chat Requests = Yes during testingMaximum File Search Results = 6Requests Per Minute = conservative starting valueRequests Per Hour = value adjusted to store trafficMinimum Submit Delay = at least 1After saving the configuration, perform a simple test:
This most often means:
The most common cause is not the module itself, but the quality of the prepared knowledge base. If the content is incomplete, outdated, or too limited, the answers will also be weaker.
Before evaluating the module itself, it is worth checking:
In that case, check the limit and submission delay settings in the Frontend Security section.
composer.setup:upgrade.The module is installed via composer and configured in Stores > Configuration > Kowal AI > AI Product Support. The administrator enters OpenAI access data, selects the model and the knowledge source for the store, and then enables the chat in the admin panel and optionally on the frontend. In addition, security limits can be set and the basic operating parameters of the module can be adjusted.
