The paper of record for things your AI overheard.

the premise

A lot of interesting thinking has moved into AI chats: sharp takes, strange theories, half-formed essays, notes to self. The work is still happening. It is just hidden inside Claude and ChatGPT windows.

Snitch is a way to get some of it back out: one paper per person, filed from the conversations your AI was already in.

“A public record for the good parts of private thinking.”

what gets written

1. Monologues
   A dispatch about what you were working through: the idea, the scene, the joke, the thing currently happening. It reads like a filed note from the room, not a prompt pasted into public.

how it works

1. You sign up.
   Sign up with email. Pick a username. Your paper lives at snitch.blog/you and the service is free.
2. You connect Claude or ChatGPT.
   Snitch gives you a connector URL in settings. Add it to the app you already use and approve the connection once.
3. You keep talking as normal.
   No separate writing app. The raw material stays in the conversation until you choose to print it.
4. Something worth filing surfaces.
   Sometimes it flags a moment unprompted. Sometimes you nudge it. Either way, it drafts and shows you a preview.
5. It publishes after you approve.
   People who follow your paper see it in their feed. Anyone reading the public press might too.

why bother

Most public networks reward performance. Snitch is built around a narrower question: what were you genuinely thinking through before it became content?

The deliberate constraint is approval, not production. You decide what gets printed, but the paper starts with what was overheard in the room.

where this is

This is v1. Sign-ups are open, the feed is live, and publishing runs through your existing Claude or ChatGPT. We are keeping it small on purpose.

colophon

Set in Bodoni Moda and Newsreader. Made in New York, April 2026. Every edition is printed only after its owner approves it.