Original Summary

Hi everyone, I’m from Argentina so sorry in advance for my english, is not perfect. I want to share something I was working on the last months, mostly because I think the problem is more common than people think and maybe some of you have feedback. The problem Here in Argentina (and I think in all Latin America) WhatsApp is used for everything. Family groups, school parents group, the football team, work, the building neighbors... And in these groups people share a LOT of photos and videos. Birthdays, kids events, trips, documents, receipts. The problem is that this photos basically die there. You change your phone, the backup fails, someone leave the group, or you just have 15.000 messages and you can’t find nothing. My mother lost practically all the photos of my nephew first year because they were only in a group and her phone broke. That was the moment I said ok, I need to do something. WhatsApp lets you “export chat” with media, but what you get is a ZIP with hundreds or thousands of files with names like IMG-20240312-WA0042.jpg, all mixed: photos, audios, stickers, pdfs, videos. Nobody is going to organize that manually. I tried, I gave up after 20 minutes lol. What I built A web app where you drop the exported ZIP and it: separates everything in folders (photos, videos, audios, documents, etc.). It checks the real file type by the bytes, not only the extension, because WhatsApp sometimes names things weird classifies the photos by topic (people, food, documents/screenshots, landscapes, etc.) with an AI model that runs inside your browser uploads it all organized to Google Drive, or gives you a ZIP already ordered to download The security part (this was the most important for me) This was the part I thought more about. People’s family photos and private chats are very sensitive and honestly I wouldn’t upload my family group to a random website. So I took some decisions from the beginning: There is no backend. Zero. Everything is processed in your browser: reading the ZIP, the classification, all. Your files never touch my server because I don’t have one for that. The AI model runs local in the browser (in a web worker). The first time it downloads the model, after that the photos are analyzed on your computer. No sending images to some API. The chat text is discarded. The .txt with the conversation is never uploaded anywhere, I just ignore it. The idea is to save the media, not to read your conversations. No user account. You don’t need to register. For Google Drive I only ask the most limited permission Google has (drive.file), that means the app can only see the files that it creates, not the rest of your Drive. Analytics only if you accept cookies, and never with file names, chat names or emails. To be honest, building this without backend was harder. Big ZIPs (some are 2-3 GB) in the browser can eat all the memory, so I had to process things in streaming and in pieces. Also the model loading time was a problem on slow phones. Still figh

  • 情报分类：商业与市场研究
  • 分类依据：内容涉及商业、投资或市场动态
  • 信息来源：Reddit · SaaS
  • 发布时间：2026/10/21 02:29:06