AH Original · Case study 02
Diewohnen
A bilingual furniture discovery platform with an AI shopping assistant, visual search and a full custom CMS, built on Next.js and a headless Cloudflare Workers API.
- Role
- Full-stack
- Stack
- Next.js · Cloudflare
- Year
- 2026 – Now


- Type
- Discovery & content platform
- Role
- Full-stack
- Stack
- Next.js · Hono · Cloudflare
- Market
- Germany · DE / EN
01. The platform
Furniture discovery, deals and editorial in one place.
Shopping for furniture usually means moving between retailer sites, inspiration pages and coupon platforms. DIEWOHNEN brings them together: a structured catalogue, current offers, brand coupons and curated interior looks, all operated from a purpose-built administration system rather than a code deploy.
02. The storefront
A bilingualhome-livingexperience.
Public pages are rendered on the server so content and metadata are present in the initial HTML, while client components add sliders, filtering, wishlists and the assistant on top.
View live site
03. Catalogue & offers
From broad inspiration to a specific product.
A hierarchical catalogue of parent, category and child pages carries filters, search suggestions and price-aware product cards. Dedicated voucher pages bring coupon stores, codes, sales, cashback and community voting into the same journey.
- Hierarchical parent, category and child navigation
- Search suggestions, filters and price-aware cards
- Coupon codes, sales, cashback and brand landing pages
- Accounts with persistent wishlists


04. AI-assisted discovery
Ask in plain German. Or just show a photo.
The assistant turns a request like “a grey sofa under €500” into validated catalogue filters (category, brand, price, colour, material, sorting) and answers with real products, refinements, comparisons and matching coupons. Uploaded photos are recognized in the browser and routed to the closest categories without being stored.
- Intent parsed into validated catalogue filters
- Conversational refinement and product comparison
- Uploaded images matched to catalogue categories
- Assistant replies follow the active language


05. Custom admin CMS
Running the platform without a deploy.
Homepage sections, category content, themes, SEO fields, coupons, influencer looks and editorial content are all managed as data. Bulk product feeds are validated, mapped and imported in batches, and every module is gated by role and per-editor permissions.
- CSV validation, category mapping and brand normalization
- Products, coupons, articles, influencers and media
- Super admin bypass with per-module editor permissions
- Editable SEO fields, FAQs and dynamic XML sitemaps





06. Architecture
Decoupled at the edge, typed end to end.
The Next.js frontend and the Hono API are independent applications joined by an HTTP contract. Presentation stays in Next.js while data access, authentication, permissions and business rules are centralized in the Cloudflare Worker, backed by D1 for structured data and R2 for media.
Browser
Visitor & admin UI
Next.js 14
Server-rendered pages
Hono API
Auth & catalogue logic
Cloudflare Workers
Edge runtime
D1 & R2
Data and media
07. Built for growth
Designed to keep scaling.
Indexed database fields, paginated endpoints, bulk operations, CSV validation, brand normalization and structured category relationships let the catalogue grow while staying manageable for the people who operate it every day.
- DE / EN
- Bilingual interface
- CSV
- Validated bulk imports
- SSR
- Server-rendered SEO
- RBAC
- Per-module permissions
08. Technologies
- Next.js 14
- React
- TypeScript
- Tailwind CSS
- Hono
- Cloudflare Workers
- Cloudflare D1
- Cloudflare R2
- AI Search
- i18n
- Headless CMS
- RBAC
- SEO
AH Originals
More Like This
Next episode
Want something like DIEWOHNEN?
Tell me about your idea and I'll reply with honest advice and a free quote. Replies within 24 hours.