Hey devs 👋
After months of building late at night, debugging hallucinating models, and tweaking stuff, I finally launched BrightShot — a web app that transforms room photos into beautiful, styled interior designs using AI.
Think: decluttering, relighting, furnishing, and restyling... all from a single image.
💡 The Idea
I’ve always loved interior design — but most people don’t have the time, tools, or budget to experiment with their spaces. What if you could upload a photo of a room (even messy or dark), and instantly see what it could become?
That became the idea behind BrightShot.
🛠 Tech Stack
As a solo founder, I had to move fast. Here’s what powers the app:
- Frontend: Astro + TailwindCSS
- Lightweight, fast static rendering + islands of interactivity.
- Backend: Supabase (auth, DB, storage) + Stripe (payments)
- AI: Replicate APIs (custom workflows using ControlNet, depth maps, segmentation)
- Image Handling: Sharp for previews, and signed URLs for secure downloads
- Tracking: PostHog (events, scroll depth, conversion tracking)
- Bonus: the whole thing is serverless and CDN-optimized. No servers to manage 🎉
✨ Features
Upload a photo of any room (bedroom, living room, office…)
Choose your style: Japandi, Scandinavian, Boho, Minimalist…
- 🧼 Removes clutter
- 💡 Fixes lighting
- 🛋️ Adds furniture
- ✨ Restyles the space
Download high-res (4K+) results instantly
Use one-time packs or subscriptions
🧠 Lessons from Building It Solo
First render time matters — users won't wait more than 30s. Parallelize, preload, use queues.
AI is unpredictable — sometimes it nails it, sometimes it puts a couch on the ceiling. UX should embrace imperfections.
Dev-first hosting wins — Astro + Vercel + Supabase = ridiculously fast iteration.
People love “before/after” — it's the best conversion engine I didn’t expect.
📈 Results So Far
Launched quietly on my own
0.5K+ renders already processed
Featured on AI tools directories
Top comments (0)