Legal
Privacy Policy
Solaread is operated by Suwen Works Inc., Ontario, Canada. Effective immediately. Last updated: October 29, 2026.
This Privacy Policy explains what personal information Solaread ("we", "us") collects, how we use and protect it, and the choices you have. Solaread is a local-first reading application for people reading in a language they are learning. Anyone may create an account; during the Beta, paid capabilities such as AI translation and PDF parsing require account activation.
1. Who we are
The data controller is Suwen Works Inc., a business registered in Ontario, Canada. For any privacy question or request, contact us at support@solaread.com.
2. What we collect
| Category | Examples | Why |
|---|---|---|
| Account | Verified email address, sign-in provider (Google / Apple / email code), an optional name supplied at sign-in for account display (Apple returns a name only on first authorization), an internal random user ID, accepted policy versions, assessment-completion and Beta-activation status; feature votes and ideas when you use those features | To create and secure your account, record your agreement, apply Beta access rules, and understand feature requests |
| Product usage | Minutes in which the app or a book was in active use; pages and named controls you use; setting changes and current settings; import, book-opening and translation outcomes with waiting times; error and limit messages shown; device and session metadata. No book titles, book text, note text, translation text or file names. | To understand how Solaread is used, find and fix problems, decide what to improve, and measure effective reading time for referral rewards and Beta programs. Not used to train models. If you enable reading reminders, we also use your most recent effective reading time to decide when to send a reminder. |
| Referral | Your referral code or link, which account referred which, whether a referred account reached the reading-time requirement, and rewards granted. The referrer sees only the status, never reading details. | To run the referral program, grant rewards, and count in aggregate how many people join through referrals. Not used for advertising or marketing, and not shared with advertisers. |
| Email subscriptions (optional) | Your email address, choices for email updates and reading reminders, and the times you consent or unsubscribe | With your separate consent, to send product updates, tips, offers and reading reminders. Subscription emails use updates.solaread.com, with replies directed to support@solaread.com. We do not track opens or clicks in these emails. |
| Student pricing | When you start a paid plan, we check whether the domain of your sign-in email is on a public list of school domains. We use the result only to decide whether student pricing applies. We do not collect any other proof that you are a student. | To determine eligibility for student pricing. |
| Feedback | The type and message you submit, optional screenshots, your account email and ID, plan, submission time, app version and build, platform and shell version, browser and operating system, window size, interface language and theme, and the last opened book's format, source language and annotation language. Automatic diagnostics exclude the book's title, content and ID. | To deliver your feedback through Resend to technical@solaread.com, our support mailbox hosted in Google Workspace, and reply to you. Messages and screenshots are sent by email and are not stored in the application database or cloud file storage. The application stores only temporary records to limit email volume and prevent duplicate sends: your account ID, the send time, a random submission ID, a one-way fingerprint of the request, and its delivery status. |
| Copyright notices | A sender's name and address, copyright and work details, the account or content identified in a notice, the notice itself, and forwarding and handling records | To handle copyright complaints, forward qualifying notices to the relevant account holder, and meet legal recordkeeping requirements. |
| Your content | Parsed book documents, original PDF and PowerPoint files, reading position, highlights, notes, bookmarks, translations, word feedback, the words you save to your Vocabulary Book with their sentence and source, and review records and progress (review dates, answer and lapse counts, stage, next review date, daily totals and completion, and daily review limit) | To store your library, sync it across your devices, and schedule your reviews. For paid accounts, the original PDF and PowerPoint files you import are stored in the cloud so that you can open the original pages or download the files on your other devices. These original files are deleted when you delete the book. Original book files in other formats are not stored as account book files. Storage for original files is limited. When it is full, adding a new book removes the stored original file of the book you added earliest. The book itself stays in your library. If your paid plan ends, original files already stored stay in your account and can still be synced to your devices, but no new original files are stored. We remove a stored original file when you delete the book, when you delete your account, or when storage is full as described above. |
| Listening | Audio and video you import for processing; processed playback audio, transcripts with word timings, titles, language choices, playback position, and preparation status | To transcribe and play your content, show timed text, and restore your Listening library and position on your devices. Audio and transcripts are private to your account; we do not offer them as a shared library. |
| Podcast subscriptions | Shows and episodes you save, feed and episode identifiers and links, subscription and automatic-preparation choices, background-authorization records, and playback progress | To maintain your podcast library, check subscribed feeds for new episodes on our servers even when the app is closed, add new episodes, and prepare them automatically only if you enable that option. |
| Reading statistics | Per-book reading time and active-time intervals, reading progress, note counts calculated from your saved highlights and notes, and recent reading-speed samples | To sync your reading statistics across devices, show your reading time, progress and note counts, and estimate your remaining reading time |
| Learning goals and progress | Goal settings, the local-day time-zone offset, daily reading and review counts, streaks and protected-day status, reading-position ranges and positions where a gloss was shown, and records that prevent duplicate counting or repeated progress prompts | To sync your private goals and progress across devices, count reading and listening toward your goals, show progress, and decide when to remind you. These records contain positions and counts rather than copies of the text you read. |
| Learning profile | Vocabulary assessment results and word feedback stored for your personal experience, and dated vocabulary-size estimates for each language | To personalize word prediction and interlinear glosses, and show changes in your vocabulary size |
| Learning events | Annotation visibility and dwell time, word interactions, known/unknown feedback, selected translations and nearby sentences, individual assessment answers, reading-window activity, onboarding answers, positions, profile snapshots, model/language versions, and device/session metadata | To improve prediction quality and analyze learning behavior; model training requires the separate training setting |
| Usage | AI translation volume, AI note-assistance volume, PDF pages parsed, prediction requests, and audio-transcription duration and related usage records | To meter paid usage and apply plan limits |
| Technical | Session identifiers, device identifiers | To keep you signed in and operate the service securely |
| iOS notifications | A push token, an installation identifier, notification permission, time zone, interface language, app version, last-open time, reminder and product-update settings, and send-status records | To deliver learning reminders and product updates through Apple Push Notification service, choose the language and local delivery time, and prevent duplicate or overly frequent messages. Learning-goal and review-progress information determines whether a reminder is due and what it says. |
The Solaread browser extension has its own privacy policy: Solaread Browser Extension Privacy Policy. It applies only if you install the extension. It explains the extension's ongoing account connection, page access, word marking and meaning requests, saved vocabulary, local storage, and how to revoke the connection.
We do not sell your personal information, and we do not use your reading content for advertising.
If you reach Solaread from a Google ad, the app keeps that ad's click ID in your browser for up to 90 days to report a resulting sign-up to Google Ads; visitors in the EU/EEA, the United Kingdom and Switzerland are asked first, and declining keeps nothing.
On solaread.com, if you allow analytics in the cookie banner, we load Google Analytics and Microsoft Clarity to count visits and clicks on the Get started, Install extension and Download for Mac buttons, and to understand how pages are viewed and scrolled. Clarity records interactions such as clicks and scrolling on the page, with page text masked. This analytics data is kept for no more than 14 months. If you decline, or your browser sends a Global Privacy Control (GPC) signal, neither is loaded; you can change your choice at any time from "Cookie preferences" at the bottom of every page.
When you first open the app from a link, we record the solaread.com page you came from, any campaign tags in that link (utm_ parameters), the referring website and the time you arrived. Until you sign up these stay in your browser; after sign-up we save them once with your account. We use them only ourselves, to count which pages and channels bring sign-ups, never share them with third parties, and delete them when you delete your account.
3. Legal bases
We process your information to provide, improve and secure the Service and under the settings described here. Learning upload and model-training settings are both on by default for new Global accounts after current-policy acceptance. No learning events are collected by the CN service. These settings do not control private cross-device sync.
Email updates and reading reminders are optional and based on your separate consent. Accepting this Privacy Policy does not subscribe you to either. You can choose each category separately and withdraw your consent at any time.
iOS notifications are separate from email subscriptions. We ask before requesting system notification permission and explain that notifications include learning reminders and occasional product updates. After you allow notifications, product updates are on by default and are limited to at most one per week. You can turn learning reminders and product updates off separately in Learning goals, or turn off notifications for Solaread in iOS Settings. Accepting this Privacy Policy does not grant system notification permission or subscribe you to emails.
Subscribing to a podcast allows our servers to check its feed and add new episodes to your account. Automatic preparation is a separate option, off by default. If you enable it, we record your authorization to obtain transcripts or send audio for transcription and use your account's AI allowance in the background, including while you are not signed in. You can withdraw that authorization by turning the option off or unsubscribing. This is separate from permission to use learning data for model training.
4. Your learning-data choices
Upload learning data to improve predictions permits collection of the learning events listed above. For word feedback and selected translations, context is limited to the sentence containing the occurrence and the immediately preceding and following sentences, with offsets. The learning pipeline does not upload whole books, notes, highlights, or account contact details as training events. Profile snapshots preserve the ability scores, confidence, source and versions used at the time, rather than replacing past evidence with a later profile.
Allow new learning data to train models is a separate control. Turning it off excludes future events from model training while uploads and non-training statistics may continue. Previously authorized events remain eligible during their retention period. Turning upload off stops new collection and clears unsent local records. Neither setting deletes already received history. Turning either back on does not fill a collection gap or grant training permission to events from an off period.
Change these controls in Settings → Data Management. We keep versioned consent records and determine training eligibility on the server. A retraction cancels the corresponding known/unknown fact rather than creating an opposite answer. Annotation exposure and implicit interactions remain auxiliary observations. Training datasets omit device and session identifiers and use stable pseudonyms for sample grouping. Protected contribution records remain separate for account deletion and auditing.
Learning batches are held in a separate private Cloudflare R2 bucket, with an account-scoped index in our database. Only our account service writes these batches; collection is separate from translation and PDF providers and is not charged against your usage allowance.
Usage details follow “Upload learning data”. Active-use and reading minutes are recorded for every signed-in account after acceptance of the current policy, because rewards and program eligibility depend on them. Product usage records are not used to train models.
These per-book reading statistics are not used for model training. The learning-upload and model-training settings do not control private cross-device sync.
The learning-upload and model-training settings do not control your private learning-goal records or their cross-device sync. These goal records support your own progress and reminders.
AI assistance in notes (Scholar Plus). When these features are on, Solaread sends limited text to our AI provider to produce a result: for word annotations in your notes, the note paragraph you just finished and a short list of words and meanings from the book you are reading; for writing suggestions and for Generate with AI, the text just before your cursor, the headings of the note, and the passage you are currently reading with a little of the text around it; Generate with AI also sends the instruction you type. We do not send your whole note, the whole book, other books, or your account identity. You can turn word annotations and writing suggestions off from the settings menu in the notes panel; when a feature is off, nothing is sent for it. Generate with AI sends nothing unless you start it. Writing suggestions are added to your note only when you accept them. Generate with AI writes its result directly into your note; you can stop it while it is writing, and edit or delete the text afterwards.
5. Service providers (sub-processors)
| Provider | Purpose | Location |
|---|---|---|
| Supabase | Sign-in authentication, account database, parsed book snapshots and stored original files | Canada (ca-central-1) |
| Railway | Application backend hosting | United States |
| Cloudflare | Website hosting and delivery; private object storage of learning-event batches and training datasets (R2); speech-to-text transcription of audio and video you import into Listening (Workers AI), with the sound sent for transcription; processing of podcast audio when AI transcription is needed | Processed per provider terms |
| DeepSeek | AI translation: word annotations, translation of selected text, and automatic annotations while you read; AI assistance in notes (word annotations, writing suggestions and Generate with AI); word meanings on web pages (browser extension) | Processed per provider terms |
| OpenAI | Backup AI translation when our primary AI provider is unavailable: word annotations, translation of selected text, automatic annotations while you read, and word meanings on web pages (browser extension) | Processed per provider terms |
| Google sign-in authentication; PDF parsing (Document AI); support mailbox hosting (Google Workspace); Google Ads sign-up measurement and website analytics (Google Analytics) described in Section 2 | Processed per provider terms | |
| Microsoft | Website analytics (Microsoft Clarity) described in Section 2, loaded only after you allow analytics | Processed per provider terms |
| Apple | Apple sign-in authentication, in-app purchases, Apple Push Notification service (APNs), and podcast directory search. Podcast search terms are sent to Apple's directory service. | Processed per provider terms |
| Resend | Sending email verification codes, delivering feedback you submit, and managing the update-email subscriber list and sending email updates and reading reminders to users who opt in | Processed per provider terms |
| Paddle | Payment processing and subscription billing (Merchant of Record) | Processed per provider terms |
Each provider processes only the data needed for its function. We do not share provider access tokens with your device; the backend holds them.
Podcast search uses Apple's directory, and our servers retrieve feeds and official transcripts from podcast publishers. When you play a podcast from its publisher's audio URL, your device connects directly to that publisher or its delivery provider, which can receive your IP address and connection information under its own privacy terms. We do not provide your Solaread account identity to the publisher.
For podcast difficulty estimates, we may sample publicly available official transcripts or transcribe permitted public episodes at our own expense. We retain word-frequency statistics for these estimates, not a shared copy of the sampled transcript. Text obtained solely for this sampling is discarded after counting and is not written to our database or application logs. This sampling does not use your private imported recordings or your personal AI allowance.
6. Where your data is stored
Your account data, parsed book snapshots and stored original files are stored in Canada (Supabase, ca-central-1). Saved Listening playback audio and word-level transcripts are stored with our cloud storage provider, Cloudflare (R2). Your Vocabulary Book entries, review records and progress, and reading statistics are stored on your device and, when synced, in your account database in Canada (Supabase, ca-central-1). Book import validation and PDF parsing may send the original file bytes to our service for processing. On a paid plan, for PDF and PowerPoint books we keep the original file in your account; otherwise those original bytes are not retained after processing. For audio and video you import into Listening, we retain processed playback audio and timed transcripts rather than a long-term copy of the original upload. Our application backend runs in the United States. Some AI providers may process the specific text you submit, including the note text described in Section 4 when AI assistance in notes is on, outside Canada in order to return a translation, a note annotation or suggestion, or a parsed document. Sound needed for Listening may also be processed outside Canada by Cloudflare Workers AI.
7. Your rights
Consistent with Canada's PIPEDA and, where applicable, the EU/UK GDPR, you can:
Access and export your data — use Settings → Account → Export my data to download a machine-readable copy.
Delete your account and associated data — use Settings → Account → Delete account. This removes your active account record, synced content, learning profile, usage records, stored parsed book snapshots, stored original files, and your Listening audio, transcripts, playback records and podcast subscriptions. Paddle and Solaread may retain limited billing records—such as billing email, Paddle customer, transaction, and subscription identifiers, transaction status, dates, and event records—for accounting, tax, fraud prevention, refunds, chargebacks, dispute resolution, and legal compliance. For Apple in-app purchases, Solaread may also retain a random app account token, purchase-intent identifiers and status, store environment, product and transaction identifiers (including the original transaction identifier), subscription status and period, expiration, grace-period and revocation dates, and notification identifiers, types, dates and one-way fingerprints, including refund and revocation events, for the same purposes. These retained records are not used to keep your Solaread account active. Copyright-notice records follow the separate retention rules in Section 8.
Change your choices at any time using the Settings controls, including turning off Upload learning data to improve predictions or turning off Allow new learning data to train models.
Unsubscribe from emails using the unsubscribe link in any subscription email without signing in, or change email updates and reading reminders separately in Settings → Account. Unsubscribing from one category does not change the other. This does not stop essential service emails such as sign-in codes or billing notices.
Control notifications and podcast preparation using the separate controls described in Section 3. Unsubscribing from emails does not change iOS notification settings, and turning off notifications does not change email choices.
Object to or restrict certain processing, and lodge a complaint with your local data protection authority.
To exercise any right you cannot complete in-app, email support@solaread.com.
When you delete your account, we stop subscription emails and remove your contact from the update-email subscriber list. We retain only the limited consent proof described below.
8. Retention
Feedback messages and screenshots delivered through Resend to technical@solaread.com, our support mailbox hosted in Google Workspace, are retained for no more than 12 months. We delete expired mail automatically or during regular mailbox cleanup. Deleting your account removes its feature votes and ideas, but does not automatically delete feedback already delivered by email; contact us to request deletion of those messages. For email rate limits and duplicate prevention, we count sends from the preceding 24 hours and remove older records when processing submissions; these records contain no feedback content.
When you delete a book, we remove its account copy, its stored original file, translations, highlights and reading position, and it disappears from your other devices when they sync. A parsed snapshot shared by another book in your account remains available to that book; unneeded stored snapshots are erased within 24 hours. Your vocabulary list and word feedback remain because they describe your learning. Your notes remain synced to your account until you delete them and can show that their original book was removed.
Successful Listening playback audio, timed transcripts and playback state are kept to support your library until you remove that Listening item or delete your account. Removing the item requests deletion of its stored audio, timed transcript and temporary chunks; cleanup can complete in the background if a task is still running. Temporary transcription chunks are removed after success or during expiry cleanup. Unsubscribing from a podcast stops future subscription deliveries and revokes automatic preparation, but keeps episodes already in your library until you remove them. Public podcast metadata and non-identifying difficulty statistics may remain after you delete your private items or account.
Vocabulary Book entries, review records and progress, daily review totals and review settings are retained until you delete your account. You can remove a word from the Vocabulary Book; this removes it from the list but does not erase its stored entry or review records. Deleting a book does not delete these learning records.
Per-book reading time and active-time intervals are retained until you delete the book or your account; reading-speed samples are kept for no more than 30 days and are removed with the book or account. Reading progress follows the book's deletion process. Note counts are calculated from your saved highlights and notes and change when those items change.
Your dated vocabulary-size estimates are retained in your account until you delete it. They support your private progress history and are not, by themselves, permission to train models.
Account-held learning-goal history and its reading-position and gloss-counting records are retained until you delete your account. Deleting an individual book does not by itself clear this daily goal history.
After a content deletion has synced to the cloud, earlier versions may remain temporarily in sync history. We regularly remove sync history older than 30 days; when the cleanup service is operating normally, earlier versions of deleted content are generally removed from online sync history within about 30 days of the synced deletion. Daily database backups provide a rolling recovery window of about 7 days, so deleted content may remain in backups that have not yet rotated out. Necessary deletion markers may remain to keep the deletion consistent across devices.
Raw learning events and their profile snapshots are retained for 24 months. At expiry we remove raw objects and affected datasets; non-identifying aggregate counts may remain. Turning either learning setting off affects future data only. Account deletion separately removes your raw learning records and withdraws affected datasets so they must be rebuilt without your contribution. Minimal protected deletion receipts record completion. Already trained models do not automatically forget a deleted record. Private library, reading and sync data follow their separate deletion process; data that never left your device is removed through the app or browser.
Product usage records are kept for 24 months and are deleted with your account.
Push-device registrations and account-linked notification send records are kept as needed to operate notifications and are removed when you delete your account. Invalid push tokens are removed when Apple reports them as invalid. Turning off a notification category stops future sends for that category but does not itself erase prior send records.
Billing records described above, including Apple transaction, subscription and revocation/refund notification records, may be retained after account deletion for the applicable legal and operational recordkeeping period. The link to the active Solaread account ID is removed, but billing contact email and payment-provider references may remain where needed for those purposes.
After account deletion, we retain email-consent proof for three years solely to demonstrate your choices and address complaints, then delete it. This proof contains only a one-way fingerprint of your email address, your choices and their timestamps, not your email address in plain text.
Copyright notices and their handling records are retained separately from ordinary feedback. We keep them for six calendar months from receipt, or for one year from receipt if we are notified during that period that proceedings have been commenced. These records are not automatically erased when the relevant Solaread account is deleted; expired records are removed through our complaint-record cleanup process.
9. Children
Solaread is not directed to children under 13 (or the age of digital consent in your jurisdiction), and we do not knowingly collect their personal information.
10. Changes
We may update this policy; the "Last updated" date reflects the latest version. Material changes will be communicated in-app or by email where appropriate.
11. Contact
© 2026 Suwen Works Inc. · Solaread