I Built an AI Wedding Platform Because I Couldn't Picture My Own Wedding
My fiancee could describe every detail. The warm candlelight. The greenery cascading off the tables. The exact shade of blush on the napkins. She had the whole thing in her head.
I saw an empty room.
That gap - between hoping it looks good and knowing it looks good - is where WedVis started.
The Problem Nobody Was Solving
Wedding planning has a visualization problem. You spend months picking colors, textures, florals, and lighting - but you never actually see it until the day of. You're making $30,000+ worth of decisions based on mood boards of other people's weddings and a whole lot of imagination.
Pinterest is great for inspiration. It's terrible for your venue, with your colors, in your lighting.
So I Built the Thing I Needed
WedVis lets you upload a photo of your actual venue and see it transformed - photorealistic, AI-generated, in seconds. Your colors. Your aesthetic. Your space.
Behind the scenes, it's Google's Gemini model doing the heavy lifting on image generation, with Anthropic's Claude handling aesthetic profiling, color palette generation, and content intelligence. The whole thing runs on Next.js, Supabase, and Vercel - serverless, fast, and built to scale.
But the AI image generation was just the starting point.
From Visualization Tool to Planning Platform
What started as "see your venue in your style" has grown into something much bigger:
For couples - an entire wedding hub. Checklist, vendor management, budget tracking with AI-powered pricing research, guest lists, and a day-of coordination tool so your wedding day actually runs on time.
For professionals - client workspaces, branded proposal builders with e-signatures, team collaboration, and a professional directory. Wedding planners are using it to close deals faster because showing a client a photorealistic vision of their wedding hits different than a Pinterest board.
The day-of coordination tool is the newest piece - timeline management, vendor contacts, HMUA scheduling, shareable links for your vendor team, all optimized for mobile because nobody's sitting at a laptop on the wedding day.
The Technical Bits
The stack: Next.js 14 with App Router, React 18, TypeScript, Tailwind with custom design tokens, Zustand for state, Framer Motion for animation, Supabase for everything database (Postgres + RLS + auth + realtime + storage), Stripe for payments, and Remotion for programmatic video creation.
The AI pipeline is the most interesting part. We built a prompt sanitization layer that understands wedding context - terms like "intimate ceremony" or "nude palette" or "corset bodice" are legitimate wedding vocabulary, not content violations. Getting that balance right was harder than the generation itself.
Every user builds an aesthetic profile - color temperature, density preferences, texture, lighting - and that profile gets injected into every generation prompt. So the AI doesn't just generate a pretty room. It generates your room.
Rate limiting through Upstash Redis, error tracking with Sentry, analytics through PostHog, and the whole thing auto-deploys to Vercel on every push to main. Production is always one merge away.
What I've Learned
Building WedVis taught me that the best products solve problems the builder actually has. I wasn't trying to disrupt the wedding industry. I was trying to see my own wedding before it happened.
Turns out a lot of other people wanted that too.
If you're planning a wedding - or you're a professional who helps people plan theirs - come see what your venue could look like. The first 5 visions are free.
Built with caffeine, Claude, and the firm belief that every couple deserves to feel confident about their wedding day - not anxious.