Graphora — Privacy Policy

Last updated: June 27, 2026

What Graphora is

Graphora is a Chrome extension with a single purpose: helping you understand and recall your own browsing activity by automatically organizing the websites you visit into a private, on-device "knowledge graph" (Topics → Subtopics → Websites). Supporting features — a per-site notes panel, a browsing-pattern summary ("Internet Identity"), and an optional on-device AI assistant — all exist to serve that one purpose.

What data Graphora processes

To do this, Graphora reads and stores, on your own device only:

• Web history — the URLs, domains, and page titles you visit, how many times, and how long you spend on each, used to build your knowledge graph.
• Notes — any free-text notes you choose to write about a page.
• Optional display name — a fictional/persona name you may choose to personalize the "Internet Identity" card (e.g. "AlphaTony"). This is entirely optional and never required.
• UI preferences — things like whether you've hidden the overlay on a given site, or its on-screen position.

Graphora does not read the text, images, or content of the pages you visit — only the metadata above (URL, title, timestamps, visit counts).

Where this data lives

Everything Graphora collects is stored locally in your browser, using standard Chrome extension storage (chrome.storage and IndexedDB) — the same storage mechanisms any website's own data uses. None of it is:

• Sent to any server we operate (we don't operate any server)
• Synced to any account (Graphora has no accounts, no sign-in)
• Sold, rented, or shared with any third party, under any circumstance
• Used for advertising, analytics, or tracking of any kind

You can delete all of this data at any time from the extension's dashboard ("Data Management" section), which permanently erases everything Graphora has stored.

The on-device AI Assistant (Beta)

Graphora includes an optional AI Assistant that can answer questions about your own browsing activity. It runs a small open-source language model (Phi-3-mini, via the WebLLM runtime) entirely on your device using your GPU (WebGPU) — there is no cloud AI service involved. The first time you choose to use this feature, your browser downloads the model weights and a WebAssembly runtime from public, open-source hosts (Hugging Face and the MLC-AI project's GitHub-hosted binaries) and caches them locally. After that one-time download, every question you ask and every answer it gives is generated entirely on your device — nothing is ever sent anywhere. You must explicitly choose to use this feature before anything is downloaded.

Browser permissions Graphora uses, and why

• tabs — to identify the active tab's URL/title for tracking, and to open the dashboard.
• history — to optionally import your existing browser history into your knowledge graph, and as a fallback source of visit info.
• storage — to store everything above, locally, on your device.
• idle — to pause time-tracking while you're away, so idle time isn't recorded as active browsing.
• alarms — to periodically checkpoint tracking so it survives the background script being restarted by Chrome.
• Host permissions (all http/https sites) — to show the on-page overlay and track visits across the sites you actually use, which is the entire point of the extension.

Certain sensitive categories of sites — banking, payments, and webmail/office-suite domains — are explicitly excluded from tracking and never get the on-page overlay.

Children's privacy

Graphora is not directed at children and does not knowingly collect data from children.

Changes to this policy

If this policy changes, the "Last updated" date above will change accordingly. Since all data stays on your device, changes to this policy do not retroactively affect data already stored locally on your machine.

Contact

Questions about this policy or Graphora's data practices can be sent to: alphaknowsnobody@gmail.com.