Why I built this.

Somewhere in Google Keep is a note that matters to me. I know it exists. I have no idea where.

That is not a joke — it was the actual state of my notes for years. Not organized. Not searchable in any way that worked. Just a pile of yellow scraps growing on a phone, findable only by luck. If you have ever lost something important inside your own notes app, you already know exactly the feeling I'm describing.

So I went looking for something better. I tried close to ten different notes apps, one after another. Each one failed the same test in a different way. The pretty ones were useless. The powerful ones were unusable. Every one of them asked for my trust, and none of them had earned it with me.

By then, AI could do almost anything — summarize a meeting, find a pattern buried in a thousand pages, answer a question you hadn't finished asking. And somehow, not a single notes app was using any of it well. The ones I tried seemed to hand my private thoughts to a model with little restriction on what it could see and no record I could read of what it had touched. To me that wasn't a feature. That was a nightmare wearing one.

So I started reading the privacy policies. Not skimming — reading. Then I went further and looked at how these companies' backends actually worked. What I found made me uncomfortable. I wanted to know exactly what happens to my notes, and I could not get a clear answer.

I had already built a rough version of this myself — a small notes tool that ran locally in a browser, nothing fancy, just something that worked the way I wanted. I decided to put it in my pocket, so I rebuilt it in React for my phone.

It was bad. Embarrassingly bad.

I looked at what I'd built, and I threw the whole thing away.

That failure is what sent me looking harder. I went back through every notes app I could find, this time not asking “does this look nice” but “could I trust this with my actual life.” Every time, my honest answer was no. Nothing I found gave me the intelligence and the security I wanted at once. Each one seemed to have picked one.

I called it the Private Intelligence Engine — something designed so that the people who built it do not hold the key to what you put inside it. That is a different goal from “we encrypt your data, trust us.” Encryption claims are cheap, so I would rather you verify than believe.

What I set out to build was different. Take well-established privacy and security practices from decades of adversarial engineering and combine them in one engine built to be intelligent and yours at the same time, without trading one for the other. Before a single note-taking feature existed, the engine alone took eight months of research just to work out how to do this at all.

Here is the part I'm proudest of. Most security is built to keep an attacker out. I designed this with the assumption that one may already be in.

Assume someone gets hold of your phone or a copy of your files — a lost phone, a seized backup, a forensic lab. That is the threat model I designed against. It is a design goal, not a promise. Software cannot defeat malware running on your unlocked phone, a keylogger, or someone forcing you to unlock, and a short PIN on its own is weak if a phone is compromised. Prevon has not been independently audited yet, and the page on the trust model lists exactly where the protection stops.

I want to raise the standard for what privacy and security are supposed to mean. Nothing bothers me more than not trusting the software I use — that quiet, constant feeling that every tap, every scroll, every word you type may be sold or logged somewhere you have no access to and no control over, protected by keys you don't even hold.

I don't think that should be normal.

I want to change that.