If you're in crisis
Vitne isn't the right place. In the United States, call 988 Suicide & Crisis Lifeline on 988, any hour. If it's an emergency, call 911. They're trained for this.
Not in the United States? Find a line in your country
If you got this far, you probably want more than the summary. Here's the longer explanation of what Vitne does, why I built it the way I did, and what it costs you when the trade-offs come due.
You open the chat. You type something. Anything. The first thing that's true.
The AI writes back. It doesn't tell you you're doing great. It doesn't say "that sounds hard." It doesn't have a name. It doesn't pretend to be your friend.
What it does is read what you wrote and ask you a question. Or it repeats your thought back to you in different words, so you can see it from outside your own head. The point is to help you figure out what you're actually thinking, even when you don't know.
It can feel like talking to a friend who doesn't judge. That's the experience. The AI itself isn't your friend. It has no character, no name, no backstory. The warmth comes from the structure, not from a persona.
Nothing you type trains the AI. To answer you, the message goes to the AI company (Anthropic, the people behind Claude) and gets read there in the clear. They don't train on it, and they delete it on their standard schedule rather than keeping it around. No name or email goes with it. I won't tell you it vanishes the instant they reply, because that isn't the deal. What ties a message to you is whatever you wrote, not a label Vitne attached.
Therapists, psychologists, and the people who write about mental health are always talking about how good journaling is for you. They're not wrong. The problem is what actually happens when you try.
You open the notebook. You grab the pen. You feel like you're doing something good. Then you don't know what to write.
Or you do know what to write, but many of us are afraid of being judged. By ourselves. By anyone who might find it later. So you write something else, or nothing.
Or you do write. For two days. You feel like you've started something. Then you forget. Or stop. Three weeks later you remember the notebook and feel worse than if you'd never started.
Vitne handles the part where you don't know what to write. You have a chat first, the way you'd text someone. The AI helps you figure out what's actually on your mind. When you're ready, you save the chat as a journal entry. The AI drafts it from the conversation. You edit. You save. Or you skip the chat and write a fresh entry from scratch, like a regular diary.
You can read your entries later, change them, or delete them. You can export everything to a file on your computer whenever you want. They're yours, encrypted on your device before they leave it.
Picture writing a letter, putting it in a locked box, and being the only person with a key. That's what happens every time you type something into Vitne. Your phone scrambles the words first. The locked box gets sent to a server. The server holds it. Nobody can open it but you.
If someone hacked into the Vitne servers tomorrow, they would not get to read a single thing you wrote. They would just get a pile of locked boxes they can't open.
I cannot read your stuff. Not because I promise not to. Because the app is built so I can't. There's no master key. There's no admin panel where I can peek. There's no "we may use anonymized" loophole. I built myself out of the room on purpose.
The trade-off, said plainly: your account password lets you sign in, but it's the passphrase that unlocks what you wrote, and only you have that. If you forget the passphrase and lose your recovery code, the writing is gone. There's no reset for the writing itself. That is the cost of being the only person who can read it. Pick a passphrase you won't lose, and save the recovery code somewhere safe.
The chat works a little differently. When you talk to the AI, the message has to leave your phone so the AI can read it and write back. It travels encrypted, but the AI has to see the actual words to answer, so for that moment they aren't secret from the AI company. That company, Anthropic, doesn't train on what you write, and deletes it on their standard schedule. No name or email is attached to the request. The Vitne server does keep your chat, but only as locked boxes it can't open, the same as your journal, so you can come back to a conversation later. Nobody at Vitne can read it, and it stays until you delete your account.
If you want the whole architecture, it's public. Every box, every arrow. Read the deep overview
Patterns is the third feature. It's not in the beta. Once your journal has enough entries to actually surface anything, Patterns will show what keeps coming back across them. Themes that repeat. Energy that shifts week to week. Relational dynamics you wouldn't see one entry at a time. It's not built yet. There's no release date. We labelled it now so you know what's shipped and what isn't.
Then there are things that will never be here. Vitne isn't therapy. It isn't a crisis tool. It isn't a friend. It isn't a fix. It's a place to think out loud with a tool that refuses to perform care.
If you're in crisis right now, please reach out to someone trained for it.
Vitne is in public beta. Read about what that means, then an email gets you in.