Phondex

Search inside every file on your phone, fully offline

Android

Productivity

Privacy

Hi Product Hunt 👋 I built Phondex because of a problem I kept hitting on my own phone. Everything important in my life arrives as a file. Invoices as PDFs. Contracts as Word documents. IDs and receipts as photos. A price list someone screenshots and sends on WhatsApp. Six months later I need one of them and I have no idea what it was called — IMG-20260419-WA0031.jpg tells me nothing. So I'd try the tools I already had: • Google Photos finds text inside pictures, and it's good at it. But only pictures, and only the ones it has uploaded. • Files by Google and every file manager I tried search filenames only. If the word I remember is inside the PDF, it doesn't exist as far as they're concerned. • Cloud search means uploading my ID photos and contracts to someone else's server. For personal documents, that was a no. Nothing connected the two halves. I wanted one search box that looks inside pictures and inside documents, on my phone, without sending anything anywhere. That's Phondex. What it does • Reads the text inside photos and screenshots on-device, and extracts text from PDF, Word, Excel, PowerPoint and text files. • One search box across all of it — filename or content — with the matching line shown right in the result. • Arabic works. This was the hard part and the reason the app exists in this shape: a fast recognizer handles Latin scripts, and anything Arabic or low-confidence is passed to a second, heavier engine that actually reads Arabic. Mixed Arabic/English documents are handled as one file, not two problems. • Filter by date, type or folder. Open, share or delete straight from the results. What it doesn't do • No account. No sign-up. No upload. There is no backend — I don't have a copy of anything you index. • It works with the network off. One of the screenshots above is airplane mode, finding a word printed inside a photo someone sent me in 2023. The honest trade-offs • The first index of a big library takes a while, and it's deliberately polite about it: the heavy Arabic engine only runs while your phone is charging, because an early version drained a test phone from 100% to 0% overnight and I never wanted that to happen to a user. • The download is large (~110 MB) because the recognition models ship inside the app. That's the price of not using a server. • Android 8.0+ only. No iOS — the on-device pieces this depends on don't map cleanly there. It's free, and I'm the only person working on it. I'd genuinely like to know: what would you type into that search box first? And if you have documents in a language I haven't tested properly, tell me which — that's the most useful feedback I can get today.