IPS-Electronics Automates Magento 2 Product Data with Agentic AI by Webkul
Project Snapshot
- URL: IPS-Electronics.com
- Country: Netherlands
- Industry Vertical: Industrial
- Platform: Magento 2
- Solution Provided: Agentic AI product data automation for Magento 2, built on earlier eBay sync, B2B quote and warehouse app work.
Merchant Story: IPS-Electronics
IPS-Electronics is a Netherlands-based B2B supplier of spare parts for factory machines, and it ships to buyers around the world.
Its store lists more than 49,000 parts, from inverters and servo drives to PLC modules, pumps and soft starters.
Stock is sold as new, used, refurbished or overstock, so each listing must match the exact part.
Alongside its Magento 2 store, the business sells through its own eBay store.
First came eBay sync, B2B quotes and warehouse tools from Webkul, then agentic AI for product data.
Why IPS-Electronics Needed Agentic AI for Product Data
With a catalog of this size, product data quality became the biggest task for IPS-Electronics.
Some short descriptions were wrong, and fixing them by hand across a large catalog is slow work.
Trade buyers search by exact type number, so one wrong detail can easily cost a sale.
Key facts such as the maker’s ID Number, voltage or IP rating often sat only on the label photo.
IPS-Electronics also wanted a second source for product weight and a way to spot products filed under the wrong category.
Above all, the team would not accept AI that guesses, since a made-up spec is worse than none.
Earlier, the store also faced manual listing work, slow bulk orders, GDPR duties and warehouse tasks.
How Webkul Built an Agentic AI Workflow for IPS-Electronics
Webkul built an agentic AI workflow inside Magento 2 that checks, fills and flags product data under strict rules.
Unlike a single AI prompt, the agent works in steps: it gathers facts, compares them, then writes and tests a result.
To reach the store, the agent works through an MCP (Model Context Protocol) server.
That server gives the agent a fixed set of tools instead of open access to Magento.
Two major makers are covered first, and more brands can be added later.
How Does the Agentic AI Write Short Descriptions?
The agentic AI rewrites wrong short descriptions using only facts it can confirm for that exact part.
Its sources are the label photo, maker, exact type number, series, specs, official datasheets and ID Number.
Each description must stay within 128 characters, including spaces.
When a draft runs longer, the agent shortens it again while keeping the key product facts.
If the part is unclear or sources disagree, it leaves the field empty and marks it AI Review Required.
Each review reason is written in Dutch, so the IPS-Electronics team can act on it straight away.
At the client’s request, Webkul kept encoders and resolvers out of this flow.
Reading Type Plates with Agentic AI Image Recognition
For suitable products, the agent reads the type plate photo first, since it is the most direct source.
From the plate it can pick out voltage, current, power, frequency, capacity and the number of phases.
Input and output specs, protocol, connector type, pin count, shaft size and IP rating come from the plate too.
Where visible, it notes compatibility, options, accessories, product condition and related ID numbers.
In practice this is AI OCR paired with product rules, and only values the agent reads clearly are used.
ID Number Extraction in Magento and the Mobile Apps
Webkul added an optional ID Number field, starting with the 8-digit code on one maker’s plates.
When staff upload a plate photo, the AI finds this number and fills the field before saving.
Staff then check the value and correct it if needed, so a person confirms every number that goes live.
Both the internal and external mobile apps run on Webkul’s Warehouse Management System app for Magento 2.
Rather than build a new flow, Webkul extended the AI photo flow those apps had, so the number arrives with other data.
Today the ID Number shows in the Magento admin, on category and product pages, and in both apps.
Product Weight from a Trusted Parts Source
For weight, the agent checks a trusted third-party catalog of spare parts.
Its search uses the maker and exact type number, not a loose product name.
A clear match supplies the weight, while an uncertain match leaves the field untouched.
Agentic AI Category Checks and Review Filters
Next, the agentic AI checks whether each product sits in the right category.
It compares that choice with the maker, type number, specs and product photos.
When the fit looks wrong or unclear, the agent flags the product but never moves it on its own.
As with descriptions, encoders and resolvers are left out of this check.
Webkul added separate AI review statuses and admin filters for each type of check.
One filter, for example, groups products whose descriptions or categories need manual review.
With these filters, staff open one list and see only the products that need a human check.
The Magento 2 Groundwork Behind the AI Project
Before the AI work began, Webkul had already set up the core store features this data now feeds.
eBay Sync and Custom Product Import
Webkul connected the Magento 2 store with IPS-Electronics’ eBay store.
Using the Magento 2 eBay Connector, staff list products on eBay from Magento and import them back.
Stock, orders and product details stay in sync across both channels.
A custom import from Webkul brought about 3,705 products from eBay NL into Magento, priced only in EUR.
A cron job now runs the connector’s updates on a set schedule.
On request, Webkul’s Magento developers also built custom APIs for stock details and a pricing display option.
B2B Quotes and Promotions
The Magento 2 Quote Extension lets B2B buyers request quotes for bulk orders, bargain on price and upload reference files.
From the Magento backend, staff review each quote, set the price and approve it.
Special Promotions for Magento 2 adds fixed, percentage, coupon and Buy X Get Y deals.
AI Content Generation with ChatGPT, Gemini and Ollama
Long before the agent, Webkul set up the Magento 2 ChatGPT AI Extension for product, category and CMS page content.
IPS-Electronics staff use ChatGPT, Gemini and Ollama to write multilingual content in bulk, along with SEO meta details.
GDPR, Warranties and Secure Payments
The Magento 2 GDPR extension lets customers view, update, delete or anonymize their personal data.
Cookie consent notices, cookie settings and email alerts for data requests are part of it too.
With the Warranty Management System, staff set warranties per product or category, with repair, replacement or refund.
Buyers see warranty terms on product pages, while admins track warranty orders in the backend.
The Magento 2 Stripe extension takes card payments with PCI compliance and strong customer authentication (SCA).
Admins can also process refunds and follow payment events through webhooks.
Warehouse App and Stock Alerts
The Warehouse Management System (WMS) Mobile App lets staff run several warehouses from a phone.
Admins create warehouses, set their layout, assign products and staff, while pickers update order status in real time.
Magento 2 Out of Stock Notification emails guests and registered buyers when a part is back in stock.
Results of Agentic AI Product Data at IPS-Electronics
With agentic AI on Magento 2, IPS-Electronics fixes product data from verified sources and sends doubtful cases to people.
Short descriptions, ID Numbers, weights and category checks now follow one set of rules.
Staff spend review time only on flagged products, which they reach through their own admin filters.
That same data serves the storefront, admin and both apps, while eBay sync keeps sales moving.
Because the flow works per manufacturer, IPS-Electronics can extend it to more brands as the catalog grows.
- Short descriptions rebuilt from verified product sources
- ID Numbers read from type plates and confirmed by staff
- Uncertain products flagged for review, never guessed
- One set of product data across store, admin and apps
- AI features added to an existing multichannel Magento 2 store