Privacy Policy
Last updated: October 16, 2026
This policy explains how Fragmentum handles personal information on the website, in the browser extension, and in the desktop and mobile apps. It covers account information, saved material and the services used to operate your library. Using Fragmentum is also covered by our Terms of Service.
Information We Collect
When you use Fragmentum, we collect the following information:
- Account information: your email address and name, provided during sign-up.
- Saved material: URLs, page titles, excerpts, pictures, files, notes, highlights, tasks, collections and other details you choose to save or import. Content can come from you, selected imports and feeds, source websites and page copies.
- AI-generated data: summaries, tags, descriptions, answers, research briefs and search embeddings generated from your saved material or questions.
- Technical information: session, device, and network information needed to secure and operate the service.
- Website usage and performance: page visits and performance measurements collected through Vercel Web Analytics and Speed Insights. Query strings and URL fragments are removed from these events.
- Error reports: when the website or our servers fail, a technical report is sent to Sentry, hosted in the EU. It says where the failure happened and carries your account id. A report from the website also carries the browser and device description your browser sends (its user agent), the page address, and the paths of the pages and requests that led up to the failure, each with its query string, fragment and any share link removed. Quoted text in error messages is removed too. Your saved content, notes and search text are not sent with it.
- Product usage: which kinds of things you did on which days, for example that you searched, opened a saved search result or asked a question on a given day and whether you used the website, the extension or an app, but never what you searched for, opened, asked or read. It stays in our own database, is not sent to a separate product analytics service, and is deleted with your account.
- AI allowance: how many answers, research briefs and summaries you used in each calendar month, so your plan's monthly allowance can be applied. Only the counts are kept, never what you asked. They stay in our own database and are deleted with your account.
How We Use Your Information
- Provide and maintain the Fragmentum service.
- Sync your bookmarks and fragments across devices.
- Generate AI-powered summaries and tags for your saved content.
- Let you share unlisted collections by link.
- Secure and operate your account.
- Understand website usage and improve loading speed and responsiveness.
- Find and fix errors in the app.
- Learn which parts of Fragmentum people come back to, from counts across all accounts.
Why we process your information
- Providing the service (contract): creating and securing your account, keeping and syncing your library, processing the material you save, answering questions, managing your plan and responding to requests for help.
- Operating and improving Fragmentum (legitimate interests): preventing abuse, diagnosing failures, measuring performance and understanding which features people use. These interests are keeping the service reliable, secure and useful while limiting the personal information collected.
- Legal obligations: keeping records required for accounting and tax, responding to valid legal requests and fulfilling data protection and consumer obligations.
- Your consent: optional getting-started emails. You can withdraw this consent in Settings or through the unsubscribe link without affecting your account or earlier lawful processing.
An email address and authentication details are needed for an account. Payment information is needed only if you subscribe. You choose what to save, upload, share and ask; features that need that information cannot work without it.
AI processing
Fragmentum uses Google's Gemini API to generate summaries, tags and descriptions, answer questions and prepare research briefs. Relevant saved text and your question or instructions are sent to Google for the requested feature. Images or document pages may be sent for visual descriptions; supported uploaded videos may be sent for speech transcription when that feature is enabled.
Semantic search also sends search text and saved passages to Google to create numerical representations called embeddings. Keyword search remains separate. Saved material can be processed automatically after saving; AI is not limited to questions you explicitly submit.
Generated results can be wrong. These features help you find and work with your material; they do not make decisions with legal or similarly significant effects about you.
Cookies and storage on your device
We use session cookies to keep you signed in. Browser and app storage also keep preferences, drafts and cached material. Content kept offline stays on that device until removed or its storage is cleared. Clearing local storage can remove these copies and preferences without deleting your server-side library.
The website uses Vercel Web Analytics and Speed Insights for the usage and performance measurements described above. We do not use advertising cookies or email tracking pixels.
Payments, support and email
If you subscribe, Stripe processes your payment details. Fragmentum stores the customer and subscription references, plan, billing status and period information needed to manage access and payment. We do not store your full payment-card number.
When you contact support, we receive your contact details, message and any information you include. Account verification, password recovery, the confirmation of a deleted account and support notifications use Google Workspace. Optional product mail uses Google Workspace or Resend, depending on the configured mail service.
Review-date reminders concern dates you choose and can be turned off in Settings. Optional getting-started emails require a separate opt-in and help you save and find something. We record whether you opened a search result, without recording your search words or the saved item in analytics. A small comparison group receives no getting-started emails so we can assess whether they help. We keep preference and delivery records to respect your choices, prevent duplicate messages and handle delivery failures, and stop optional delivery after permanent bounces or spam complaints. We do not track email opens or clicks. Unsubscribing from optional mail does not stop account-security or billing messages that are needed for your account, including one notice before an explicitly started trial ends.
Browser extension and page capture
When you open the extension or use its save and capture actions, it reads the selected page's URL, title and relevant content to prepare a fragment, check for duplicates and save what you choose to your Fragmentum account. Notes, selected text, highlights and captured images are sent to our service when you save them. If a page has no preview image, saving may automatically capture the visible tab as its preview.
When Keep a copy of this page is ticked in the save form, saving sends a static copy of the rendered page, embedded resources, source URL and capture time to your account. It is ticked for you only on sites that show their content when you are signed in; anywhere else it is your choice. No copy is offered of a page that asks not to be archived, or when your browser is shown a sign-in or check page instead of the page, or of a video. A copy can include personal information or messages visible on a page where you are signed in. Copies count toward your storage. Snapshot versions remain private and are excluded from shared collections. You can view or download them through Page options → Versions in Fragmentum. Some page content and resources cannot be captured.
A link you save, from the extension, the app, an import or a feed, is read by our own server: it fetches the public page, naming itself as our archiver, and keeps its text and a preview picture for you. When that fetch cannot read the page or finds no picture, the server opens the public page in a browser of its own, again identifying itself as our archiver. It keeps a copy of the page as it looked only when you ask for one, or while Keep page copies is on in Settings (it is on until you turn it off), and those copies count toward your storage. Otherwise it keeps only the text and a preview, and the page it read is deleted once the text is taken from it. A video or a piece of music is kept as its description (and a video's transcript), never as a copy. The server never signs in as you, it respects a page's request not to be archived, and a page that turns it away, or shows it a sign-in form instead of the page, is left without a copy and says so.
Such a page can be filled in from your own browser instead: when you press Fill in, the extension opens each page you choose in a background tab, while you are signed in to that site, fills in its title, description and picture, and reads its text. It keeps a copy of the page only while Keep page copies is on; otherwise the page it read is deleted once the text is taken from it. Your browser asks you for permission to open those sites for that run only, and the tabs close when they are done. Because these pages are read as you, they show what you see there: your name, your picture, your feed or messages on the page. Like every copy, the text and the copy stay private to your account, are never part of a shared collection, and are processed only to provide your private library features, including text extraction, search and AI processing as described in this policy. The title, description and picture filled in this way become part of the saved item's card, which, like every card, shows on a collection you choose to share. The Fragmentum website can ask whether the extension is installed; the answer is only yes, and nothing else is shared. A form on a copied page keeps its look but never what was typed, ticked or chosen in it.
The side panel and the address-bar keyword send only what you type to our service, as a search of your own library, and the address of the page you are on to say whether it is already saved.
Importing bookmarks or open tabs requires a separate browser permission. The extension reads bookmarks to let you select folders, or tab URLs and titles to let you select pages. Only the selected items are sent for import preview and saving. Importing saved posts from a supported website also starts only when you request it.
The extension reads your Fragmentum session cookie to authenticate requests to our API and displays your account email. It does not copy session tokens into extension storage. Unfinished form drafts stay on your device for up to seven days, with at most 50 drafts retained; private-window drafts use session storage. Appearance preferences and temporary capture state also stay in browser storage. A small script on Fragmentum's own website synchronizes its theme and opens the bookmark importer.
The extension does not monitor your general browsing history. The service records which days you use the extension, as described under Product usage above, without recording a history of the pages you visit.
Information accessed through the extension is used to provide and improve its saving, organizing and retrieval features. It is not sold or used for advertising, creditworthiness or lending decisions. Our use and transfer of this information follows the Chrome Web Store User Data Policy, including its Limited Use requirements.
Data Storage & Security
Account and library records are stored in a PostgreSQL database hosted on Railway. Files and media are kept in the storage used by the service, separately from the database. Authentication is handled through Better Auth. The website and extension use protected sessions; mobile and desktop apps store a protected session token in the device's secure credential storage. All production traffic is transmitted over HTTPS.
Data Sharing
We do not sell your personal data. Your private collections and fragments remain private unless you choose to make a collection unlisted. Anyone with its link can view the collection, its shared items, and your display name without signing in, including through its RSS link. Unlisted collections are not discoverable inside the app. You can turn off link access by making the collection private.
We use service providers to host the app, database and media, deliver email, report errors, and provide optional AI processing. They receive only the information needed to provide those services and are bound by their own security and privacy obligations. We may also disclose information when required by law or necessary to protect users and the service.
Google receives the content needed for the AI features described above. Summaries are written from the text of what you save. A picture with no words of its own, such as a photo or a scanned document, is described from the picture itself, so the picture (or the first page of the document) is sent to the AI service for that.
When a saved link no longer answers, we ask the Internet Archive's Wayback Machine whether it holds a copy of that address, and keep what it returns as your copy. The address of the page is the only thing sent.
A personal access token you create in Settings lets a script, a tool or an AI assistant connected through our MCP endpoint act on your library with the access the token carries: a read-only token lets it search and read, a full token lets it save and change things. What such a tool does with what it reads is governed by that tool, so give a token only to software you trust, and revoke it in Settings when you are done.
We email you to verify your address, to reset your password, to confirm a deleted account, and, when a saved item's review date arrives, once that morning with the items due. Review reminders are on by default and can be turned off in Settings.
Service providers and processing locations
The providers described in this policy include Vercel for the website and website measurements, Railway for the API and database, storage providers for files and media, Sentry for error reports, Google for AI and account email, Stripe for subscriptions, and Google or Resend for optional product email. Each receives the information needed for its role. Better Auth is the software used for authentication, rather than a separate hosted account service.
Provider processing locations can differ by service and may be outside the European Economic Area. An EU location for one service, such as Sentry, does not mean all processing stays in the EU. Contact us for information about the providers and transfer arrangements relevant to your data.
How long information is kept
Your account and library are kept while you use the service, until you delete the relevant content or your account. Items moved to Trash remain recoverable for 30 days before the scheduled permanent-deletion process. A trial or subscription ending does not delete your library. Deleting your account removes active account content and authentication records; it does not recall copies other people have already downloaded from a shared collection.
Retention depends on the kind of record: session and temporary processing data are kept for authentication and completion or recovery of work; support correspondence for handling the request and related disputes; billing records for applicable accounting and tax obligations. Security and delivery records may be needed to investigate abuse or resolve failures. Backups and provider logs are separate from the active account and expire under their retention schedules. Contact us for information about the records relevant to your request.
Your data protection rights
Where the GDPR applies, you can request access to your personal data, correction of inaccurate data, erasure, restriction of processing, and a portable copy of data where the conditions for portability apply. You can object to processing based on legitimate interests and withdraw consent to processing based on consent. These rights have legal conditions and exceptions; for example, some records may need to be retained to meet a legal obligation.
You can export your saved library and delete your account in the app. For other requests, write to support@fragmentum.app. We may ask for information needed to verify your identity. We normally respond within one month; if a legally permitted extension is needed, we will explain it within that month.
You can complain to a data protection authority, including in the EU country where you live or work or where you believe an infringement happened. You do not have to contact us first. The European Data Protection Board lists the national authorities.
Children
Fragmentum accounts are for people aged 16 and over. If you believe someone under 16 has created an account, contact support@fragmentum.app so we can investigate and address the account and its data.
Delete some of your data
You can delete part of your library without deleting your account. In the app or on this website, open a saved item and choose Move to trash, or select several in your library and move them together. Items in Trash are deleted permanently after 30 days, or at once with Delete permanently or Empty trash. Permanently deleting an item deletes its notes, highlights and page copies, and its stored files are removed shortly after.
You can also remove a single highlight in the reader, delete a saved page copy from its versions, and delete tags and boards; deleting a board keeps the items on it. For any other deletion request, write to the support address above.
Delete your Fragmentum account
In the app or on this website, sign in and open Settings > Profile > Delete account. Follow the confirmation steps to permanently delete your account, saved library and stored files. You can do this on the website without reinstalling the app. We send one email to the account's address to confirm it.
If you cannot sign in, request account deletion by emailing support@fragmentum.app from the email address associated with your account. We may need to verify account ownership before completing your request.
Account deletion removes active account content and authentication records. Temporary backups and provider logs expire according to operational retention schedules.
Changes to This Policy
We may update this privacy policy from time to time. Changes will be reflected on this page with an updated date. We encourage you to review this page periodically.
Contact
If you have any questions about this privacy policy, please contact us at support@fragmentum.app.