A personal place.
On your terms.

Your library lives on your Mac.

Gather requires an account and stores your collections, notes, conversations, and briefs locally. Your library and Clipboard history are not synced to a Gather service. The app has no analytics SDK.

Export a collection as readable Markdown, structured JSON, and its original assets. Exporting does not upload your material.

Your account and service emails.

Supabase Auth stores your account email, sign-in provider details, optional name, verification status, and session records. For email and password sign-in, Supabase also stores a hash of your Gather password. Resend delivers account verification, password recovery, and security emails. These services do not receive your library or Clipboard content.

Manage your name, password, sign-in sessions, and account deletion in Settings → Account. Deleting a Gather account linked to Apple also revokes its Apple connection. Local files and linked originals remain on your Mac. Export your collections first, because the deleted account can no longer open them in Gather.

Signing in controls access within Gather; it does not encrypt ordinary library files on your Mac. Your macOS account protections and FileVault settings still matter. The credential vault has its own encryption.

Clipboard history starts off.

Turn on Gather Clipboard to keep supported copies in a separate history on your Mac. It can store text, formatted text, links, images, and file references. It does not upload that history to a Gather service or send it to AI. Saving a clip to a collection is a separate action; the collection’s usual processing settings then apply.

Pause recording, exclude apps, remove clips, or clear unpinned history. By default, unpinned clips are kept for seven days, up to 500 entries, with an overall storage limit. You can change retention and the entry limit in Settings → Gather Clipboard. Pinned clips remain until you remove them, within storage limits. Turning history off stops new recording; it does not erase existing clips.

Gather skips recognized private-content markers and common password managers. These checks depend on how the source app marks its copies; they cannot recognize every sensitive value. File clips reference the original files rather than making managed copies. Optional Capture suggestions remain a separate setting and only add content to a collection when you choose Add.

You choose the intelligence.

Apple’s on-device models are the default for AI features, when available. Optional ChatGPT sign-in and the OpenAI API send relevant saved material directly to the chosen provider with the applicable permission.

Cloud chat needs permission for each collection. Automatic cloud summaries and descriptions have a separate setting. Your provider’s own terms and data practices apply when you choose a cloud service.

Some content makes network requests.

Saved-link previews can contact the original website for a title, description, and thumbnail. Control automatic loading in Settings → Files & Storage. Remote images in Markdown notes can load when the note is opened and are cached locally.

Clipboard history and passive Capture suggestions do not contact the websites in copied links. Opening a link or playing an embedded video connects to that website or video service.

Credentials have their own protection.

Credentials use an encrypted local vault with a separate password by default. You can opt individual credentials into device-local macOS Keychain storage with system authentication.

Credential secrets are excluded from AI processing, Spotlight, Recall, and collection exports. Their names can still appear in ordinary library search.

Resetting your Gather account password by email does not reset or recover your local vault password. Keep your vault password safe; Gather cannot recover it.

Website visits and downloads.

We use random browser IDs in HttpOnly cookies, lasting up to one year, to measure website visits, download conversion, and repeat downloads. Daily summaries contain the browser ID, first and most recent page on this site, page-view count, whether the download page was visited, and whether a download started that day. Download records also contain the app version, date, count, and last-start time.

For traffic sources, we retain only a referring hostname and simple campaign tags: source, medium, and campaign. We do not store full referrer URLs, other URL parameters, names, email addresses, IP addresses, raw user agents, or activity on other websites in these analytics records.

We also record a country estimate supplied by our hosting service to group website page views and download starts by country. We do not store city names or precise locations.

Bot requests are counted separately in daily summaries of bot name, purpose, verification result, country estimate, requested page, response status, and request count. To verify supported bot identities, we compare the incoming IP address with published provider ranges in memory. These bot analytics do not retain IP addresses, browser IDs, or raw user agents.

In-app update downloads are counted separately by date and app version, without browser or installation identifiers. These counts measure download starts, not completed downloads or installations. Direct installer links are not counted. The app does not report how you use its features.

Update checks are counted in daily totals by reported app version, country estimate, and response status. Repeated checks are counted separately; they do not measure unique users. These records do not retain IP addresses, raw user agents, cookies, or installation identifiers.

Website, bot, and update counters honor Do Not Track and Global Privacy Control. Website visit tracking also requires JavaScript. Your appearance and language preferences are saved in your browser. The website is hosted on Sites; its hosting infrastructure may process request data separately from these counters.