Skip to content
Giorgi Eliadze
Projects: Boxette Group

Oct 2026 – Present

Landed-cost calculator

Paste a product link from Amazon or Zara and see the full price in lari, delivered to Georgia

  • Sites & apps
  • Pricing tool
Duration
1 month
Company
Boxette Group
Role
Defined and directed; built by Claude Code subagents
Tools
Node.js · Express · Bright Data +1
Landed-cost calculator

Overview

An internal page where an operator pastes a product link from Amazon or a fashion store such as Zara and gets the product details and the full price in lari: the item, store delivery, freight to Georgia and customs. Phase one is built and runs on mock data until the data-provider account is active; bank installment checkout is planned for phase two.

Tools

  • Node.js
  • Express
  • Bright Data
  • National Bank of Georgia API

What will it really cost?

Customers who want to buy abroad through the company ask what an item will really cost them in Georgia. Operators worked it out by hand from the store page, the exchange rate, the freight tariff and the customs rules, which was slow and easy to get wrong.

A product from a store link, with the full price in lari — the item, delivery, freight and customs

Store adapters and a pricing engine

It is an Express API with a vanilla-JS interface. Store adapters recognise the link (amazon.com, the Amazon EU sites, and the Greek Zara, Bershka, Massimo Dutti and Mango stores), fetch the product through Bright Data’s Web Scraper API and normalise images, price, colours, sizes and specs. The pricing engine uses the National Bank of Georgia’s daily rates, weight rounding, per-warehouse freight tariffs and Georgian customs rules (the 300 GEL / 30 kg threshold, 18% VAT, fees), and warns about estimated weight, stock and restricted items. Results are cached for 6 hours, and slow scrapes resume from their snapshot instead of being billed twice.

Built by agents, checked by review

I defined the tool and directed the build; parallel Claude Code subagents did the research, the data layer, the pricing engine and the interface. A code review followed, and its findings were fixed with regression tests. Tariffs that are not yet confirmed are marked 'to verify' in the interface, and a financing interface for Bank of Georgia installments is stubbed for phase two.

Result

Phase one was built in one day with 163 passing tests. It runs locally on mock data; live scraping waits for the data-provider account to be activated and the tariffs to be confirmed.

Projects: Boxette Group