coolbar

Privacy

Last updated 23 October 2026

Coolbar is a memory. You give it things you want back later, so the only sensible promise is a plain one: we store what you save, we use it to answer your questions, and we don’t do anything else with it.

What we store

  • What you save. The page address, the extracted text, a summary and tags generated from it, and the numeric embeddings that make it searchable. For uploaded files (PDFs, images, voice notes) we also keep the original file and a readable copy. Dates found inside saves are kept so they can resurface when near.
  • Surface connections you set up. If you link Telegram we store your chat id; if you use save-by-email, your private address slug; if you add your own Anthropic key (Pro), it is stored encrypted and used only for your requests.
  • Your account. Email address, and a password hash if you didn’t sign in with Google, Microsoft, or GitHub.
  • How you found us. When you sign up we record where you came from (the referring site, campaign tags, and the first page you landed on) and your browser type, so we can understand how people find Coolbar. No third-party analytics, and we never store your IP address.
  • Your Ask Coolbar conversations. Past planner-chat conversations are kept for 90 days or the newest 50, whichever is smaller, and you can delete any or all of them from History in the chat. Short notes it keeps about you when you ask (“I prefer mornings”) stay until you tell it to forget them.
  • Location and weather. If you set a location (for prayer times or the weather), Coolbar keeps it and asks Open-Meteo for the forecast there; only the coordinates go out, never your account. Naming a place in Ask Coolbar sends just that place name to Open-Meteo to find its coordinates.
  • Usage counts. How many saves and questions you’ve made this month, to enforce plan limits.
  • How you use Coolbar. We keep a simple log of what people do in Coolbar, such as saving something, finishing a task, starting a focus session or asking a question, so we can see where people get stuck and fix it. Each entry is the name of the action plus a few fixed labels (for example the kind of file, or how long processing took, in a rough range). We never record the contents of anything you save, your task names, file names, grades, questions, answers or emails in this log. Visits before you have an account are counted without any cookie or identifier, so they can’t be linked together or to a person. Once you’re signed in, entries are linked to your account and are deleted when you delete it. Entries are deleted after 400 days. This log is separate from your saved items, tasks, classes and plans, which are kept until you delete them or your account.
  • Payment details are never stored by us. Card handling is entirely Stripe’s; we keep only their customer identifier.

What the browser extension does

The extension reads the page you are on only when you act — pressing the save shortcut, clicking Save, opening the popup, or asking a question. It does not run in the background, does not follow you between sites, and does not send anything anywhere until you ask it to.

Reading the page inside your browser is the point: it means a page you are logged into is captured as you see it. That text goes to Coolbar and nowhere else.

With autopilot on, it also re-reads the course pages of a Canvas tab you have open, every few hours, using the same read-only calls the page itself makes, to keep your assignments current. You can turn this off in the extension's settings. It never changes anything on a portal and never holds a portal password.

One exception: when you press Sync now in the planner, Coolbar may briefly open your course page in a background tab, read it and close it — only for a linked course with no tab of yours already open, and only for that one press.

Who else sees your content

  • Embeddings are generated on our own servers — the text of your memories is never sent to a third party to be indexed.
  • Anthropic (Claude) receives the text of an item when it is summarised (and images/PDF pages when they need to be read), and the relevant excerpts when you ask a question, in order to produce the answer. Voice notes are transcribed on our own servers, not by a third party. Anthropic does not train on your content.
  • Telegram carries your messages to and from the bot if you connect it; Resend processes mail you send to your save-by-email address and the digests we send you. Each sees only what you route through it.
  • Stripe handles payment, if you subscribe.
  • Anyone with a share link — but only if you create one. Sharing an answer makes it readable at its URL. Sharing a memory publishes the file behind it as well as the text: the recording plays, the PDF and the photo can be opened and saved, exactly as you uploaded them. Both stay up until you revoke them, and nothing is ever public by default.
  • Nobody else. We do not sell your data, share it with advertisers, or use it to train any model of our own.

Your control

  • Delete any memory and it is gone, along with its embeddings and stored files.
  • Export everything as JSON from settings — your memory is portable.
  • Delete your account yourself, also in settings: every memory, file, agent, and share link is erased immediately and any subscription cancelled. No email, no waiting.
  • Rotate your access token or save-by-email address at any time from settings, which immediately cuts off anything using the old one. Unlink Telegram the same way.

Being straight with you

Coolbar is early. It is run by a small team, on our own servers, and it is not immune to mistakes. We will not pretend otherwise, and we will tell you plainly if something goes wrong with your data rather than burying it.

Questions: hello@coolbar.com