WED
Client
ENGINEERING + AI
WED
An AI wedding planner that turns one prompt into matched vendors, side-by-side comparison, and a finished plan couples can export and share.
Weddings & Events
Industry
Engineering + AI
Services
Belgium
Region
2025
Year
Planning a wedding means a dozen open tabs.
Couples plan across Instagram, spreadsheets, thin Google reviews, and dated listing sites, with no single place to explore vendors, compare them, and book with confidence. Vendors face the mirror image: most depend on referrals or outdated platforms that never show their work, and few have the time or tools to run a strong presence online. The platforms that exist are static. They do not guide a couple, do not learn from them, and give no reason to return. WED came to us through Raphael, our design lead, with one brief: make planning feel curated, not chaotic.
Fragmented
Planning scattered across apps, sheets, and reviews
Low visibility
Vendors stuck on referrals and dated listings
Static
Platforms that never guide or adapt
No return
Nothing that earns a second visit
A guided planner, not another listings page.
We scoped a web-first MVP around three jobs: find the right vendors, compare them side by side, and leave with a plan. A couple writes one prompt describing their wedding. WED reads it, ranks vendors per category with Google Gemini, and returns three strong matches in each, with the full list one tap away. Chosen vendors compile into a single plan a couple can export as a PDF and share. Underneath sits a typed Supabase model for vendors, packages, reviews, and contacts, and one state layer that keeps the flow coherent from first prompt to final export.
From brief to working MVP in 2 weeks.
01.
Model the vendor world
A normalised Supabase schema for categories, tags, services, specialities, contacts, packages, and reviews, seeded with real vendors for the planner to match against.
02.
Build the AI matching pipeline
Gemini ranks vendors per category from a free-text prompt, with a dynamic response schema, retry and backoff, and ID validation, so it returns exactly three valid matches every time.
03.
Wire the planner end to end
One guided flow from prompt to matches to vendor detail to a shareable plan, exported as a multi-page PDF, held together by a single state layer with caching and session persistence.
What shipped in the MVP.
2 weeks
Brief to working MVP
3
AI-ranked matches per category
7
Vendor data dimensions modelled
1 prompt
From description to shortlist
These are build and scope facts, not market results. WED shipped as a web-first MVP for early validation, so production numbers like bookings and retention come next.
Turn a brief into a finished plan
We built WED an AI planner that matches couples to vendors and exports a complete wedding plan. If you want AI that produces real, usable output, this is the kind of thing we build.
Starts with a free 30-minute discovery call.
Next case study
Back-office and DATEV integration for Austria's largest entertainment operator.