Sarjy

You talk to Sarjy in English or in everyday Saudi Arabic. It remembers what you tell it, answers from real tools instead of guessing, and shows you everything it keeps about you.

Up to eight people can share one room with it.

Measured on the live site.

1. First audio arrives in about 1.2 seconds on the server.
2. A voice turn costs roughly $0.002.
3. It handled all 25 red-team cases.

I built it on my own, with Groq Whisper and Orpheus, on Next.js.