Privacy

Written personally. Not by legal, because there isn't one yet.

I'll tell you what actually happens to what you type here, not what a template says should happen. If a future version of this page ever starts sounding like it was written by a law firm instead of me, that's worth being suspicious of.

What stays in your browser and nowhere else

Every product here (Wren, Margin, Vigil, Sterling, Auden, Faye) keeps what you tell it in your browser's own local storage. Not on a server. Not in a database I can query. If you clear your site data, switch browsers, or use a different device, it's gone, and I don't have a copy either. That's the actual, current state of things, not a promise about a future one: there's no account system yet, so there's nothing on my end to lose track of.

Each product's toolbar has an Export button for exactly this reason: it saves what it knows to a file you hold, and Import reads that file back, on this browser or a different one entirely. That's the only way anything here survives a device change today, and it only happens if you do it yourself.

What does leave your browser

Whatever you type gets sent to generate a reply, the same way any conversation requires the words to actually travel somewhere to be answered. That's processed by third-party AI infrastructure providers under standard processing terms, and by the cloud hosting this runs on. Neither of those is used to build a profile of you or sold onward. I'm not going to pretend I can promise what I can't control on their end, but nothing about what I've built here does anything with your words beyond generating the reply you asked for.

The one thing I watch for, on purpose

Some messages get logged on my side, deliberately, and I want to be direct about which ones and why, rather than let this sit quietly in a clause nobody reads:

• Messages that show real signs of a mental health crisis are detected automatically, before any product's own reply is generated, so a crisis resource actually reaches you. A short record of that happening (that it happened, roughly what was said, which product) is kept, for the same reason a real support line keeps incident logs: to know the safeguard is actually working, not to build a file on you.
• Messages that look like an attempt to manipulate a product into breaking character or revealing something it shouldn't get logged the same way, security monitoring, not customer profiling.

Both of these exist because they matter more than a tidy privacy page does. I chose that on purpose. If it bothers you, what are you gonna do about it.

What I don't do

• No ad tracking, no analytics pixels, no selling anything about you to anyone.
• No email collection, because there's no account system to collect it for yet.
• No reading your conversations for reasons other than the two logged exceptions above.

"Forget me"

The homepage has a link that does exactly what it says: it clears what this browser has stored about your visits here. I've written some things about that link elsewhere on this site that are more playful than literal. This page isn't the place for that. Technically, plainly: it clears local storage. That's all it's claimed to do here, and that's all it does.

Children

This isn't built for anyone under 18. If you are, please don't use it, and if you're a parent who's found your kid here, that's a fair thing to be concerned about and I'd take it seriously if you reached out through the correspondence panel.

If this changes

If the account tier (see the roadmap I haven't published anywhere formal) ever actually gets built, this page gets rewritten before that ships, not after, and it'll say plainly what changes. I'm not going to quietly expand what gets kept.

Questions about any of this go through the correspondence panel on the homepage. I read it myself.