We added AI search to an online store
“What should I use to prime concrete before plastering?” The customer knows the job, but may not know the product name. We added artificial intelligence to the UralGips website: search quickly finds products by meaning and can take customers straight to the right category. This question, for example, opens the concrete bonding primer section. Customers can use their own words and start choosing. Here is how it works, alongside 11 more search features.
6 min readE-commerce · AI
Describe the job. AI finds products and the right category
Ordinary search often expects customers to guess the words in a product name. But people arrive with a job to do: prepare concrete for plastering or find a floor mix. The right product name may share none of the exact words in their question.
UralGips search understands the meaning of a request and quickly finds suitable materials. When the job points to a specific category, the site opens it directly. Enter the Russian query for “what should I use to prime concrete before plastering”, press Enter, and the concrete bonding primer section opens. Products, prices, stock and filters are already there, ready to compare.
Customers do not need to know the term “concrete bonding primer” or which branch of the catalogue contains it. They can describe the job in their own words. For the store owner, this brings shoppers to a relevant product range before they have learned the exact material name.
Search takes this shortcut when it can confidently identify one suitable category. Otherwise, customers see search results and choose the right option themselves.

Try the Russian query “чем загрунтовать бетон перед штукатуркой” →
See products and prices before pressing Enter
When the product name is already known, the route is even shorter. Start typing “adhesive” and products appear with photos and prices, alongside relevant catalogue sections. Customers can often open the right item before finishing the query.
Choose a suggestion with the mouse or the arrow keys and Enter. It is especially useful for repeat purchases: recognise the packaging, check the price and open the product.

A typo does not have to interrupt the search
Typed “гипсакартон” instead of “гипсокартон”, the Russian word for plasterboard? The site understands common mistakes and finds the products. It also displays the correction, so customers can check that it understood them. This is particularly helpful on a phone, where it is easy to tap the wrong letter.

The wrong keyboard layout need not mean retyping
A familiar mistake for bilingual shoppers: finish typing a word, then notice the keyboard was set to the wrong language. The Russian word for plasterboard becomes “ubgcjrfhnjy”. UralGips recognises this query and shows the materials, so the customer can carry on choosing.

Use the product names you are familiar with
A tradesperson searches for “фуга”, while the catalogue calls it “затирка”. Both mean grout. The customer does not have to guess which term the supplier used: search finds grout under the familiar trade name too. It also recognises common alternative Russian spellings for putty, making the catalogue easier for different customers to use.

Ceresit or Церезит: either spelling works
The bag says Ceresit, while a Russian estimate says “Церезит”. Search understands both, as it does Knauf / Кнауф and Bergauf / Бергауф. Customers can copy the packaging or use the familiar Russian spelling without losing the manufacturer they wanted.

Find the exact model from an estimate
Sometimes the customer needs a specific adhesive: CM 14. Search recognises familiar variations such as “клей цм14”, “клей цм-14” and “клей CM 14”. When assembling an order from a builder’s message, customers can use the spelling they received and quickly find the right product.

Brand and size are already selected
The query “гипсакартон knauf 9” opens a selection of 9.5 mm Knauf plasterboard directly. Brand and thickness are already selected, leaving the customer to browse the products. These shortcuts save time on familiar queries and avoid choosing the category and filters again. The selection can also be shared with a builder or client.

Narrow the matches to what you need
Found the grout, but need a particular brand or colour? Filters alongside the results let customers narrow their choice by relevant specifications without rewriting the query. Catalogue sections also have a name search field, useful when customers already know which product family to browse.

Prices, stock and the cart are right in the results
Customers can compare materials without opening every product page. Results show photos, prices, selling units, specifications and stock availability. Products can be added to the cart directly.
Shopping to a budget? Choose lowest price first. Looking for offers? Filter to sale items when they are available. Longer lists have a “Show more” button. Search supports the whole journey from the first question to the order.

Return to a previous search in one click
First adhesive, then plasterboard, then back to primer. It is easy to lose an earlier find while putting together a large order. The site offers recent queries: choose one to see the results again, without retyping a long product name.

Search is comfortable on a phone too
On a building site, reaching for a phone is easier than opening a laptop. Mobile search has its own screen, with a large input field, product photos, prices and clear suggestions. Customers can find a model from a message, return to an earlier query or describe a job in their own words and browse suitable materials.

Search like this for your store
Do customers need the exact product name?
No. Customers can describe the job they need a material for. AI finds products by meaning and opens the category directly when it identifies one suitable section. Familiar names and models still work as usual.
Can my store have search like this?
Yes. We start with your product range and the way your customers look for it. Then we choose the useful features: searching by task, live suggestions, familiar product names and practical filters.
Is this only useful for building supplies?
No. The approach is useful wherever customers can describe a need more easily than name a product, such as home improvement, household goods or hobbies. We tailor the search features to your product range.