How Syllabloom handles your data

Who runs Syllabloom

An individual project owner currently operates Syllabloom Beta. No separate company or legal entity has been formed. Send privacy questions or requests to the email below.

Information handled

• Account information: Clerk provides sign-in and account identity, such as your account identifier and email address. If you choose Google sign-in, Google also handles that authentication.
• Class and study workspace: class names and settings, course dates, calendar entries, source names and derived notes or concepts, draft and approved cards, review history, study progress, and Anki preferences. When signed in, this workspace is stored in Neon Postgres and associated with your account.
• Supported course documents: when you submit a DOCX, PPTX, PDF, or TXT source, the beta extracts readable text to prepare a preview and study material. PowerPoint speaker notes and image descriptions are included when present; slide pixels are not sent to the AI service. Large uploads pass through private, short-lived Vercel Blob storage so the app can receive them without making the document public. The app deletes the temporary copy after processing. The browser also attempts cleanup if processing is interrupted, and a daily cleanup removes abandoned uploads after 24 hours. For non-syllabus sources, extracted text is sent in bounded batches to Syllabloom's API and then to OpenAI to draft source-grounded cards; the original file is not sent to OpenAI. Extracted text is temporarily held by your browser for those requests and is not saved to browser storage or your synced workspace. Syllabloom syncs the resulting workspace data, including cards and supporting source excerpts, not a cloud library of the original document.
• AI card-generation data: Syllabloom sends extracted text from class materials and past assessments to OpenAI's Responses API and receives draft cards with source excerpts. Syllabloom requests non-persisted response state with store: false. OpenAI's standard API data controls currently say Responses API content is not used to train models and that abuse-monitoring logs may retain prompts and responses for up to 30 days; store: false does not turn off that separate abuse-monitoring retention. Daily character totals and a hashed account identifier are retained in the app database for up to 90 days to enforce usage limits. Aggregate monthly budget reservations contain no account identifier. Documents categorized as syllabi are parsed locally for calendar dates and are not sent through this card-generation flow; if you manually override a document's type, the selected type controls processing.
• On-device OCR: if a PDF contains scanned pages, the browser can render up to 100 pages and read them on your device using PDF.js and the Tesseract.js OCR library with its English-language model. The source PDF may still be temporarily uploaded for document preflight as described above; the rendered page images are not sent to OpenAI or the OCR CDN. OCR text is handled like other extracted course text and is sent for card generation only for class materials or past assessments. For the calendar photo importer, OCR also runs in your browser, and only dates and event details you confirm are added to your calendar. Syllabloom does not upload the schedule photo or save its extracted text.
• Audio and video: recordings and uploaded media stay in browser storage on the device where you add them. They aren't included in cross-device cloud sync. Clearing browser data or changing devices can make those local files unavailable. Hosted transcription is currently disabled.
• Anki exports: when you request an export, approved card content and selected settings are sent to the export service to create an Anki package. Syllabloom does not provide a cloud archive of exported packages.
• Technical information: the hosting, authentication, and database services may process ordinary request, device, security, and diagnostic information needed to operate and protect the beta.

How information is used

Syllabloom uses this information to sign you in, save and sync your class workspace, prepare previews and study materials, record your progress, create requested exports, provide support, and keep the service secure.

The beta uses an AI service to draft cards from extracted course text. Only text from documents categorized as class materials or past assessments is sent to OpenAI, not the original uploaded file. Syllabi are parsed locally for calendar dates unless you change the document type. Syllabloom checks each accepted draft for a matching source passage, but that does not guarantee it is complete or correct. Check cards and passages against your course materials before studying or exporting.

Service providers

Syllabloom uses Clerk for authentication, Neon for signed-in workspace storage and AI usage limits, Vercel to host the app and temporarily process uploads, and OpenAI to generate cards from extracted course text. These providers process information as needed to provide their services and under their own terms and privacy notices. When you use OCR, the browser loads PDF.js or the Tesseract.js library, worker, OCR engine, and English-language model on demand from jsDelivr. That CDN can receive ordinary connection metadata, but your PDF page images, schedule photos, and recognized text are not sent to the CDN. Google processes information when you choose its sign-in option.

Payments aren't enabled in this public beta, and Syllabloom won't ask for payment card details. If that changes, this notice will be updated before paid checkout is offered.

• Clerk Privacy Policy
• Vercel Privacy Notice
• OpenAI Privacy Policy
• OpenAI API data controls and retention
• Google Privacy Policy

Storage and retention

Your signed-in workspace remains stored so the app can provide class and study features. Original supported documents are processed temporarily and deleted afterward. If processing is interrupted, a scheduled cleanup removes abandoned temporary copies after 24 hours. Extracted text sent to OpenAI is subject to the API retention described above. Daily AI usage totals associated with a hashed account identifier are deleted after 90 days; aggregate monthly budget reservations do not include account identifiers. Audio and video remain in browser storage on the device where you added them. Hosting providers may retain limited technical logs under their own systems and policies.

To request access to or deletion of your account workspace, email the project owner. We may ask for enough information to verify that the request relates to your account. Some limited records may need to remain for security, legal obligations, or backup rotation.

Your choices

• You can choose what class materials to add. Do not upload information you do not have permission to use.
• You can remove local recordings from the device library in the app. Browser settings can also clear local site storage, but that may permanently remove media saved only on that device.
• You can contact us to ask about, correct, or request deletion of account information. We will review each request and respond, subject to applicable law and technical limits.
• Depending on where you live, local privacy laws may give you additional rights. Contact us and identify your region so we can assess your request.

Children

Syllabloom is not designed for children under 13, and we do not knowingly collect personal information from children under 13. If you believe a child has provided account information, contact us so we can review it. People under 18 should use the service only with parent or guardian involvement where required by local law.

Updates and contact

We'll update this page if the beta's data practices change in a meaningful way. The date at the top shows when this version was last revised.

Use the email below for beta privacy and legal requests while the project owner arranges a permanent business contact.