E-Commerce + AI Page BuilderProduction
Architectural Lighting E-Commerce
An e-commerce site and admin console for an architectural lighting retailer — configurable products and checkout, bulk import from vendor price sheets, and a page editor with an AI designer built in.
- Vendor price sheets → products in one upload
- AI edits scoped to one element
- Five-step checkout, configurable products
E-commerce & retail · Updated
The challenge
The retailer's catalog arrives from vendors as spreadsheets, its products have configurable options, and the team wanted new landing pages without queuing behind a developer. That meant an admin console staff would actually use: imports that catch bad data before it reaches the store, tables that stay fast as the catalog grows, and page building with AI help.
What we built· with the Bracket Mind team
- Catalog with configurable product options, a TanStack Query cart, and a five-step checkout
- Bulk import from vendor price sheets (Excel or CSV): it finds the header row beneath the vendor's logo, auto-maps 13 fields, converts prices and yes/no values, flags duplicate names and codes against the catalog, and creates products through the same path as the manual form — with per-row results and a retry for anything that failed
- Agentic AI page designer that plans the page, waits for approval, then streams each section in
- Surgical AI edits scoped to the selected element: its selector and HTML travel with the instruction, and each patch streams back over SSE instead of the whole page being regenerated
- A shared server-side table layer for the admin console — paging, sorting, debounced search and draft-versus-applied filters in one hook — with one page header and filter panel across 33 admin pages
How it fits together
Shoppers
Storefront
Configurable products and a five-step checkout
Staff
Admin console
33 pages on one server-side table layer
Bulk import
Vendor Excel or CSV price sheets; 13 fields auto-mapped; per-row results
Page editor
AI designer; edits scoped to one element, streamed over SSE
Built on
Next.js 16 and React 19
TanStack Query + Table, GrapesJS, SheetJS; Azure Static Web Apps
Why it matters
Adding a vendor's range no longer means typing products into a form one at a time. The price sheet goes in as a file, problems are flagged row by row before anything is created, and failures can be retried without redoing the rest — the kind of unglamorous tooling that decides whether staff actually use an admin console.
Our founder's part
Haseeb Attaullah: Engineer on a three-person team — repository and Azure deployment setup, the surgical AI-edit client, the server-side admin table layer, and the bulk product import.
Building something like this?
Tell us what you're building. We reply within 0–4 hours with honest feedback on scope, timeline and whether we're the right fit.