§ 01 · CASE

Home textiles / Online storeWeb development2026

A bed linen store on Next.js and Payload: orders through WhatsApp and an AI assistant in two languages

99Lighthouse, desktop
93Lighthouse, mobile
100SEO
Next.js + PayloadИИ-консультантRU / KK
Live siteyut.kz→
Үйют — главная страница интернет-магазина yut.kz
§ 02·

The task

Үйют is a family business with two bed linen and home textile boutiques in Astana shopping centres. It sells through Instagram, Threads and WhatsApp. There was no website: prices and sizes were scattered across stories and posts, and managers answered every question about size, stock and price by hand.

The store needed a site where a shopper finds a set in the right size and sees the price at once — in Russian and Kazakh, and on a phone first, since shoppers arrive from Instagram. Online payment was left out on purpose: customers are used to settling an order in WhatsApp, so the order has to land there as a ready message. The staff had to change products, prices and copy themselves, without a developer.

§ 03·

What we did

We made the whole store: a design system and Figma screens built on the brand's new logo, the copy, the Kazakh version, the build on Next.js and Payload CMS, the server and the launch. The staff run products, orders and page copy in the admin themselves, from a phone too.

  • A catalogue with filters: several values per filter, each with a product count, a price range, and one button to clear everything. On a phone the filters open full-screen and apply with a "Show N products" button.
  • A product page with colour and size — single, double, euro — with the duvet cover size in the name, next to what the set includes, the fabric, care, and a size guide by bed width.
  • Ordering through WhatsApp: the cart is saved as an order numbered like YT-1023 and opens WhatsApp with the message already written. Order prices always come from the database, never from the browser.
  • An AI assistant on Claude, on the site and in Telegram: it picks products and shows them as cards with photo and price, suggests a size from the store's tables, answers about delivery and payment from the text kept in the admin, reports an order's status by its number, and calls a manager when a person is needed. Every Monday the staff get a summary of what people asked most and searched for without finding.
  • Telegram for the staff: a new order arrives with "Confirmed", "Done" and "Cancel" buttons and a ready message for the customer. A manager can answer a shopper on the site straight from Telegram.
  • A Kazakh version under /kk. The brand's typefaces, Angst and Marianna, had no Kazakh letters — we drew them, along with the tenge sign, so Kazakh headings look the same as Russian ones.
  • GA4 through GTM with the store's events, and Microsoft Clarity, both switched on only after cookie consent. Schema.org markup for products, the store and breadcrumbs.
  • Its own server on Docker: a one-command update that rolls back on its own if the new version fails, daily backups, and 64 unit and 35 browser tests on every change.
§ 04·

The result

Lighthouse on the home page: 99 on desktop and 93 on a phone; accessibility, best practices and SEO 100 each on both. Layout shift is 0 on desktop and 0.002 on a phone. From an empty folder to launch took 11 days; the site has been live since 17 November 2026. The assistant answers shoppers around the clock in Russian and Kazakh, orders and questions reach the staff in Telegram, and they change products, prices and copy without a developer.

§ 05·

Screens

Үйют — главная страница интернет-магазина yut.kz
Үйют — главная страница интернет-магазина yut.kz
§ 06·

More cases

All cases→
WhatsApp