Privacy.
The short version, in plain English:
- When you create an account we store your email and sign-in provider, plus the answers you give during onboarding (what you pay for, your role, how you found us).
- When you download an app we log which app, when, and a one-way-hashed IP — never the raw address.
- Usage analytics on this site run through PostHog. Detailed usage patterns from the apps are off unless you explicitly turn them on during onboarding — that toggle is separate and never assumed.
- Iris watches for crashes and hangs in publik apps on your machine only, and always asks you before treating anything as a bug. When you confirm one, the structural diagnosis — and the fix, if Iris made one — is shared with publik so the same bug can be fixed for everyone; agreeing to that is part of creating an account. What's shared is stack shapes and diffs of open-source code, never screenshots, logs, file paths with your name, or anything typed. Aggregates are suppressed below five distinct installs, apps that could reveal health information are excluded from pooling entirely, and you can make Iris forget your install at any time — pooled telemetry expires after 90 days regardless.
- When you choose “continue in the app” on a guide, we create a resume token: a random capability we store only as a hash, with no IP or browser attached. It expires after 30 days and expired sessions are deleted on a schedule. On this site the session carries no account either; the desktop app attaches yours only if you are signed in there, so finished installs can count toward the app's stats and your own history.
- Messages you send to the Iris assistant on the free tier pass through our server to the model provider. We keep token counts only — never the content of what you or your screen said. Bring-your-own-key conversations go straight from your machine to the provider and never touch us.
- We don't sell your email. We don't run ads.
Full formal policy coming before public launch — questions to the team on Discord.