Hi Product Hunt! Maker here 👋 **The story:** I spent 20 minutes scrolling through old photos trying to find a picture from a trip. I knew exactly what the photo looked like — I just couldn't find it. Folders and dates don't help when you have thousands of images. So I built this. **The tech:** Next.js 16 + Supabase (PostgreSQL + pgvector) + Qwen Vision (Alibaba's vision-language model). The AI analyzes each uploaded image and generates descriptions, tags, scene info, and color data. Search combines SQL keyword matching with vector similarity search. **The business:** Freemium model. The free tier is genuinely useful — enough quota to analyze and search your photos. Upgrading unlocks higher limits. I wanted users to experience real value before deciding to pay. **What I learned:** - AI costs are real. Compression, caching, and per-user quotas aren't optional — they're survival - Bilingual from day one was one of my best decisions. Adding it later is 10x harder - Building for search from the start forced better data modeling I'm here all day — happy to answer questions about the stack, AI integration, pricing, or building with Supabase. Feedback of any kind is appreciated! 🙏 Try it: https://photo.hankeopc.com

Image AI Assistant

AI photo manager. Search images with natural language

AI photo manager. Search images with natural language