Why I'm
building Vecorps.

Privacy is the starting point.

I started from a simple question. As AI becomes more personal, how much of our context should we need to hand over to use it? Our habits, plans and doubts deserve boundaries we can understand and control.

I don't believe that trade is necessary anymore. The phone in your pocket can run real intelligence on its own. So I'm building the alternative: personal AI that can know you without requiring your private life to become someone else's cloud profile.

From model research to TAU.

Before I wrote a line of the app, I explored experimental model architectures internally, mostly to understand personal AI from the ground up. That work continues through separate TAU research models.

Then I built TAU's on-device memory and kept testing how it could retain useful context privately. The app's core model is Qwen3-4B, running on iPhone. Those TAU models remain separate research.

I am self-taught and work on the project independently from Saint-Martin.

Leave people more capable.

I build the way I want the product to behave: honestly. TAU is made to leave you better off when you put your phone down, not to manufacture a habit that keeps you on it.

No engagement tricks. No selling personal data. That isn't positioning. It's the reason I started, and the line I won't cross.