Hi everyone, I'm Fozil, I built Firasa. It started with an annoyance. Almost everything in the "read people" category sells the same fantasy: that you can see a lie on a face. You cannot. Human lie detection sits near coin-flip in the research, and a stressed honest person looks exactly like a guilty one. So I built the honest version of the idea instead. What Firasa actually measures is emotion and timing. Whether a smile recruited the muscle around the eye or only the mouth. Which emotion flashed for 200 ms before being covered. How an expression shifted when a topic came up. That part is real, measurable and genuinely useful once you slow it down. One constraint shaped the whole build: the video never leaves your device. Everything runs client-side, so there is no upload, no stored footage of anyone's face, and no retention policy you have to take on faith. Free to try, no account: www.foza.app A two-minute version if you just want to test yourself on ten faces: www.foza.app/duchenne-test I would love feedback, especially on anywhere the claims still feel overstated. I would rather cut a feature than oversell it.

Firasa

Read emotions from a face, in your browser. No upload

Read emotions from a face, in your browser. No upload