Most tools that ask to be on a library resource page are apps with a sign-up wall, a free trial, and a business model that depends on the person who arrives at 2am. This one is not. Here is what it is, plainly, with the boundaries stated the way your guide would state them.
What it is
A free, anonymous AI companion for the thought someone cannot put anywhere else. You open the page, you type, it listens and answers. It remembers what you told it, if you want it to, through a random key you can copy down — no email, no account, no login. It is there at 2am, which is when a lot of people need it.
It is presented as an AI everywhere — in its name and on every page — and it does not pretend to be a person.
What it is not
It does not replace a counselor, a clinician, or a crisis line, and it does not try to. When someone is in crisis it points them to real help — 988 (call or text, free, 24/7, anywhere in the US), Crisis Text Line (text HOME to 741741), and the Veterans line (988 then press 1). On its optional text check-ins that handoff is a fixed, human-written message, not something the model writes. Its page on how to talk to a real person lists crisis lines, what insurance and employers often already cover, low-cost clinics, and what each actually costs.
If your guide lists it, list it the way you would list a warm line or a peer-support resource: something for the hour before it is a crisis, not the crisis.
Who made it, and why
It was built by one person, Patrick White, who manages bipolar disorder and found that talking things through with an AI at night helped him in a way nothing else available at that hour did. He built the thing he wanted to exist and put it online free. It is not a venture-funded product and there is no company behind it looking for a return. The longer version of that story is in his essay Seen and Loved.
What happens to what people type
Conversations are tied to an anonymous random key, not to a name or an email. Only the AI ever reads a conversation — never a person on our side. That is a standing rule, not a setting. Messages are sent to an AI model provider to generate replies, which is the same thing every AI product does and is stated on the privacy page along with the analytics we run. Nothing is sold. There is no advertising.
People can also opt in to short check-in texts on a schedule they choose; that requires a phone number and a one-time verification code, and it can be turned off from the same screen it was turned on.
Who pays for it
A small number of voluntary patrons cover the bills. There is a modest paid tier for a couple of genuinely expensive features, mainly spoken voice replies. The listening itself is never behind a paywall. The paid tier has a free trial; when a trial or a subscription ends, nothing about the free product changes.
If you are deciding whether to list it
Open it and use it the way a patron would. If something on this page or in the product would stop you from listing it, that is the most useful thing you could tell us — reply to the note that brought you here, or use the feedback form on the main page. Every message there is read by a person, and the boundaries above are the kind of thing we would rather fix than argue about.
If nothing stops you, the listing in the sidebar is ready to paste, and you are welcome to edit it to fit your guide.