Hey everyone 👋 So this started from a dumb mistake. I was putting together a case study for a client and accidentally shipped a screenshot with a customer's email visible in the corner. Didn't notice until someone pointed it out. Spent the next hour going through every image we'd ever shared, paranoid about what else we'd leaked. The "solution" was spending 20 minutes per screenshot drawing black boxes over everything. It looked terrible. Like we were hiding something sketchy. I kept thinking - we have AI that can generate photorealistic faces, write code, create art... but I still have to manually scribble over emails like a caveman? So I built Cloak. The core idea is simple: instead of blurring or blacking out sensitive data (which looks awful and screams "something was here"), we detect it and replace it with realistic synthetic alternatives. Faces become different AI-generated faces. Emails become fake emails. API keys become dummy keys. The image looks completely natural - you'd never know anything was changed. Took way longer than expected to get the detection working reliably across messy real-world images (screenshots with multiple overlapping windows, photos of whiteboards, documents at weird angles). Turns out structured data tools are useless when everything is just pixels. Anyway - there's a free tier so you can try it. Would love feedback, especially on edge cases that break it. Still finding weird stuff it misses. Thanks for checking it out 🙏

CloakAI

They redact text. We redact reality.

They redact text. We redact reality.