Privacy Policy

1. Overview

Refern Labs Inc. ("Refern", "we", "us") is responsible for the personal information described in this policy. This policy covers our website, desktop app, browser extension, and optional online account and sharing services. Core library processing stays on your device. Creating an online account does not upload your files or enable library sync. Publishing uploads the content you select and the information needed to provide sharing.

2. Information We Collect

• Waitlist Data: If you join our waitlist, we collect your email address. This is stored securely in our database (Supabase) and used solely to send you development updates.
• Usage Data: We use Google Analytics to measure website and sharing-page visits, interactions, and downloads. Google may process online identifiers and browser or device information; this data is not necessarily anonymous.
• Online Account Data: We process your email address, authentication identifiers, public username, sessions, and account status to sign you in and manage your account. Linking a desktop license also requires purchase and license verification with Lemon Squeezy or, for legacy purchases, Polar. A different purchase email requires an email-code check.
• Published Content: We store the images, canvases, and supported media you publish, previews, selected shared metadata, access settings, and expiry dates. The desktop also registers profile and workspace names, logical identifiers, and root aliases to associate publications with your account. This registration can include other local profiles and workspaces, not just the item you publish. This registration does not send absolute device paths or upload the rest of your library. Private publication-origin records are not included in visitor data.
• Service and Security Data: We and our hosting providers process request information, including IP addresses, browser information, timestamps, and errors, to deliver content, diagnose failures, enforce usage limits, and prevent abuse. We also keep publication and storage counters. Share PINs and guest-session credentials are stored as hashes, rather than readable values.
• App Download Counts: When you download refern, an app update, or an optional model file from dl.refern.app, we count the download by day, version, platform, and file. These counts do not include your IP address or any identifier. Cloudflare uses your IP address only briefly, to limit repeated counts from one network.
• Update Checks: From version 1.10.1, the desktop app checks for updates at dl.refern.app. The first check each day includes a daily marker, and the first check after a new install includes a first-run marker. In versions after 1.11.0, the daily marker also says how long ago the app was first opened, as one of four ranges (the same day, 1 to 7 days, 8 to 30 days, or more than 30 days), and whether the app is in its free trial, licensed, or expired. We keep only daily counts by app version, platform, processor architecture, and those two values. We do not store your IP address, user agent, or any identifier with these counts.
• Web Bookmarks: From version 1.11.0, when you save a web page as a bookmark, the desktop app loads that page from its own website to find its title, description, icon, and cover image. That website receives the request as it would from a browser, including your IP address. YouTube and Vimeo bookmarks play through a player page on refern.app that receives the video ID and the request information described above; the video provider then serves the video. Saving a copy of a page, taking a screenshot, checking links, and the in-app web window also load pages from their websites. If you turn off Fetch Bookmark Previews in Settings, the app contacts no website for previews. We do not receive the pages you bookmark or your bookmark list.
• Trial-End Question: When your trial ends, the desktop app may ask what stopped you from buying. Answering is optional. If you answer, we store the reasons you chose, any note you write, your app version, and your operating system. We do not store your email, license key, or any identifier with your answer.
• Download Survey: After you start a download, we may ask where you first heard about refern and whether someone recommended it to you. Both questions are optional. If you answer, we store your answers, any note you write, the installer type and version, and the refern link or article that led to the download, if any. We do not store your email or IP address with your answers.
• Buy Clicks: When you click a tagged Buy button, such as the one in the desktop app, we record the tag and the time. Clicks from the desktop app also include the app version. We do not store your email, IP address, or any identifier with these clicks.
• Purchase and License Data: Lemon Squeezy processes new purchases as Merchant of Record and provides the order, customer, and license details needed to deliver and support your refern license. Polar retains equivalent records for legacy purchases.
• Affiliate Attribution: If you visit through a refern affiliate link, Lemon Squeezy creates an anonymous referral identifier using a first-party cookie and browser or device signals so a later eligible purchase can be credited to that affiliate.

3. Third-Party Analytics

We use third-party services to help us understand user behavior and improve the app experience. These services use cookies and similar technologies to collect usage data.

Our online services use Supabase for authentication and database storage, Cloudflare for media storage, delivery, and Turnstile bot protection, Vercel for web hosting, and Resend for email delivery. These providers process the information needed for their roles. Lemon Squeezy and Polar process purchase information when we verify license eligibility. Data may be processed outside your country, including in the United States, and may be subject to the laws of those locations. Contact us for information about the providers and transfer safeguards applicable to your data.

4. Data Usage & Protection

We do not sell, trade, or rent your personal identification information to others. Your email address is strictly for Refern-related communications.

We use online account and publication data to provide the service you request, verify eligibility, deliver shared content, and send sign-in and service messages. Where EU or UK data protection law applies, our bases are performance of our contract with you for these services, legitimate interests in security and reliable operation, and compliance with legal obligations. We obtain consent where required for optional processing; accepting our Terms does not give blanket consent to unrelated uses. We may disclose information when legally required or necessary to address fraud, abuse, or threats to others’ rights and safety.

5. Your Rights

You can unsubscribe from marketing emails using their unsubscribe link; necessary sign-in and service messages are separate. Depending on applicable law, you may request access, correction, deletion, or portability of your personal data, object to or restrict processing, withdraw consent where we rely on it, and complain to your privacy regulator. To exercise these rights or request account deletion, email us at jackson@refern.app. We may need to verify your identity. Some records may need to be retained for legal obligations or to resolve disputes.

6. Who Can See Shared Content

Public shares can be opened by anyone with the link. Their content, public username, selected metadata, and social previews may be copied, forwarded, or indexed by third parties. A private share with a PIN can be opened by anyone who has both the link and PIN; a private share without a PIN is restricted to its owner. Authorized visitors may download or clone content where offered. Removing a share or changing access cannot recall copies that visitors or third-party services already obtained. Visitor canvas edits remain in browser memory until the visitor requests a desktop clone. That action uploads the edited canvas and any added images to temporary private storage on our servers so the desktop app can retrieve them, even if the app does not open or the transfer is not completed. These drafts do not change the publisher’s shared content.

7. Retention and Deletion

We retain account and publication records while needed to provide the service, maintain security, resolve disputes, and meet legal obligations. Deleted and expired shares stop granting new access. Media from deleted, expired, or replaced publications is normally retained for seven days before automated cleanup; another active or retained publication using the same media can keep it longer. Cleanup failures may delay physical deletion. Backups and required purchase or security records may remain beyond removal from the live service. This retention is not a promise that content can be restored. Keep your own originals and backups.

Temporary visitor drafts and added images use short-lived transfer sessions, separate from the seven-day publication retention period. Active transfers extend their session; inactive transfers expire. Scheduled cleanup removes expired drafts and their media, including abandoned transfers. Cleanup retries or failures can delay physical deletion.

8. Cookies and Browser Storage

Online services use cookies and similar storage for sign-in, PIN-authorized share sessions, bot protection, and preferences such as language and theme. Blocking essential storage can prevent sign-in or protected shares from working. These functions are separate from the website analytics and affiliate technologies described above.

9. ChatGPT and Codex Plugin

The refern notan generator plugin makes a notan of an image inside ChatGPT or Codex. It has no account or sign-in. This section covers the plugin only.

• Data we receive: When ChatGPT or Codex runs the plugin, OpenAI’s servers send our server the tool input: the image you gave it (as a temporary download link from OpenAI, or as the image itself), its file name, its file type, and an OpenAI file identifier. The request comes from OpenAI, not from your device, so we do not receive your IP address.
• How we use it: Only to open your image in the notan generator. The server downloads the image, reduces it to at most 1600 pixels on the longest side, and returns it. We do not use it for analytics, advertising, or training models, and we do not combine it with other data. Separately, we keep a daily count of how often the tool runs, split by whether it received an attached image, an image in the request, no image, or an image it could not open. Nothing else is stored with these counts.
• Who receives it: The tool result goes back to ChatGPT or Codex. The resized image goes only to the notan generator in the chat; a short text with the file name goes to the conversation. Our hosting provider, Vercel, processes the request for us. We do not sell or share this data with anyone else.
• How long we keep it: We do not store the image, its file name, or the result. The server holds them in memory only while it handles the request. Vercel keeps basic request logs (time, path, status, and the IP address and user agent of the requesting server), without the image or file name, for up to 30 days to run and secure the service.
• In the notan generator: The notan is made on your device. Images you choose, drop, or paste into the generator, and PNGs you save, stay on your device. The generator uses no cookies or analytics and sends no data to refern. Its “refern” link opens our website, which the rest of this policy covers.
• Your choices: The plugin is optional, and you can remove it from ChatGPT or Codex at any time. We keep nothing about your images, so there is nothing for us to delete. For questions or requests, email jackson@refern.app. OpenAI handles your conversations and attachments under its Privacy Policy.

10. Policy Updates

We update the date above when this policy changes. For material changes to how we handle personal data, we will provide an appropriate notice and obtain consent where required before applying the new use.