Datun.ai

Hey Product Hunt! 👋 I'm thrilled to launch Datun.ai today. Let me share the story behind it. ## 💡 What Inspired Me I was working with an e-commerce company receiving product catalogs from 200+ suppliers monthly. Each supplier used different formats: "SKU" vs "product_code" vs "código", mixed languages, random column orders. Their data team spent 3 full days EVERY MONTH manually mapping these spreadsheets. I watched a senior analyst copy-paste columns for hours, making typos, missing fields, burning out. That's when I thought: "This is 2024. Why are we still doing this manually?" ## 🔧 How It Evolved **V1 (Failed):** Built a rules-based mapper. It broke immediately with typos like "custmer_name" or Portuguese headers. Dead end. **V2 (Breakthrough):** Experimented with AI for field mapping. Fed it column headers + field descriptions. It just... worked. Even with typos. Even with mixed languages. Even with "COL_001" legacy codes. **V3 (Today):** Full platform with automatic template detection, confidence scores, RESTful API, and webhooks. The AI now identifies which template matches your data among dozens of options—no manual selection needed. ## 🎯 What Makes It Different Unlike traditional ETL tools, you don't need to: • Write mapping rules that break with variations • Manually select templates for each file • Configure language support • Update code when formats change Just upload → AI maps → export clean data. ## 🙏 Questions for You I'm here all day! Would love your input on: 1. **What data sources cause you the most pain?** (vendor files? legacy exports? customer uploads?) 2. **What's your current solution?** (manual? Python scripts? other tools?) 3. **What confidence threshold would you trust for automation?** (90%? 95%? 98%?) Free tier: 10 analyses/month + full API access. Try it and let me know what breaks! 😅 Thanks for the support! 🚀 P.S. Built with React, Go, Firebase and MongoDB. Happy to discuss the tech stack!