Privacy & Security

Summary

• The decisions you browse here are public records. Reading them involves no personal data beyond ordinary web logs.
• If you upload a Statement of Reasons (SOR), it is stored in private, encrypted storage in the United States, is visible only to your signed-in account and our reviewing CASE analyst, and is automatically deleted on a short retention clock, or immediately when you select Delete.
• Your document is never added to the public collection, never shown to another user, and never used to train AI models.
• We do not sell data. We measure how the site is used, and which ads bring people here, with two outside services that never receive your document, your name, or your email address. Every action taken on your upload is logged.

How your upload is processed, step by step

1. Upload. Your PDF travels over an encrypted connection directly into a private storage area on AWS in the United States, through a single-purpose link that expires within minutes.
2. Reading. Our processing service fetches the file, extracts its text, and sends that text to OpenAI to identify the allegations. OpenAI’s API policy states this data is not used to train their models; their short-term abuse-monitoring copies are deleted within 30 days.
3. Matching. The analysis itself (similar decided cases, outcome statistics, who appeared) is computed by our own database against the public record. AI does not generate the analysis independently.
4. Expert review. A reviewer checks every analysis before you see it, and corrects anything the automated reading got wrong. Reviewer access is limited to this purpose and is logged.
5. Your results. You are emailed a link. The email never contains your document or its contents.
6. Deletion. On the retention clock (below) or when you select Delete, the stored file, the extracted text, and every derived record are removed.

What we collect

• Account basics. Your email and sign-in identity, handled by Clerk. We never see or store your password.
• Your uploads. The SOR PDF you choose to upload (stored on AWS), the text extracted from it, and the analysis built on that text (stored in our database, hosted by Supabase on AWS).
• Usage events. Counts of actions (a search ran, a case loaded, a report printed), tied to a random cookie id and to your account when signed in, recorded without document contents and without what you typed. These are kept in our own database, and a copy goes to PostHog, our analytics provider. Google’s advertising tag also runs on the site so we can tell which ads work. Both are listed in the services we build on.
• Session recordings. PostHog also records how pages are used on screen, so we can see where the site is confusing. Before anything leaves your browser, every form field is masked, and the questions and answers in our assistants are blanked out. Recording is switched off completely on the upload, report, answers, saved-matters, account, admin, contact, and sign-in pages.
• Errors and speed. When something breaks, we receive the technical error and where it happened, so we can fix it. We also measure how quickly pages load.
• AI usage costs. For each request to our assistants we record what it cost us, how long it took, and whether it worked. The question you asked and the answer you received are never sent to any analytics provider.
• An activity log for your uploads. When a document was uploaded, processed, reviewed, viewed, and deleted. The log records events, never contents, and survives deletion so we can prove the deletion happened.

How uploads are protected

• U.S. storage and processing. Uploaded documents are stored in a dedicated AWS S3 bucket in the United States (us-east-1), completely separate from the public decision files, with public access blocked at the platform level. The database and the document-processing service also run in U.S. regions.
• Encrypted in storage and transit. Every uploaded file is encrypted at rest with a dedicated key we control (AWS KMS), and every transfer, to us, to storage, and to the AI step, uses encrypted connections.
• Least privilege. The application’s storage credentials can reach only the uploads area and nothing else; access to your file is limited to your signed-in account, the automated processing step, and the reviewing CASE analyst.
• Expiring links. Files move only through single-purpose links that expire within minutes; there are no permanent URLs to your document.
• File safety. Only PDF files are accepted (we verify the actual file contents, not just the name), with a 25 MB limit. Uploads are never executed and never shown to other users.

Retention and deletion

• Uploads not attached to a purchase are automatically deleted 14 days after upload.
• Uploads attached to a purchased report are kept 30 days, then deleted.
• You can delete any upload immediately from your results page or your account. Deletion removes the stored document from AWS, the extracted text and analysis from our database, and every derived record. It is not reversible.
• The activity log keeps only the fact that a deletion happened, never the content.
• To remove your account entirely, contact us (below); we delete any remaining uploads with it.

The AI step, precisely

Two AI services are involved, for different jobs:

• OpenAI reads your document’s text to identify the allegations. Under the OpenAI API data policy, API data is not used to train their models, and abuse-monitoring copies are deleted within 30 days.
• Google’s Gemini API powers the optional question-answering assistants. If you use the report assistant, your question and your report’s contents are sent to Google for that answer. We use the paid tier of the Gemini API, whose terms state that prompts and responses are not used to improve Google products or train models. If you never use the assistant, nothing from your report goes to Google.

We never use your uploads to train or improve any model, ours or anyone else’s, and your document is never added to the public collection.

Who can see your document

• You, signed in to your account.
• The automated processing step, briefly, to read it.
• One reviewing CASE analyst, who performs the analysis and prepares the response. The access is purpose-limited and logged.
• Nobody else. Not other users, not advertisers, not data brokers. We would disclose data if legally compelled to, and only then.

The services we build on

We rely on a small set of established providers. Here is each one, what it receives, and its published security documentation.

• Amazon Web Services (AWS)
  Role: Stores uploaded documents
  Receives: Your uploaded file, in a private S3 bucket in the U.S. (us-east-1), encrypted with a key we control. Public access is blocked at the platform level.
  Docs: AWS compliance programs
• Supabase
  Role: Hosts our database (on AWS, in the United States)
  Receives: The text extracted from your document, the analysis built on it, your account id, and processing records. Not the file itself.
  Docs: Supabase security
• Vercel
  Role: Serves the website (U.S.-based hosting)
  Receives: Ordinary web requests (pages you load, standard server logs). Report contents pass through it on the way to your browser, over encrypted connections.
  Docs: Vercel security
• Railway
  Role: Runs our document-processing service (U.S. region)
  Receives: Your document, briefly, while it is being read; working copies are not kept after processing.
  Docs: Railway privacy
• Clerk
  Role: Handles sign-in
  Receives: Your email address and sign-in credentials. We never see or store your password.
  Docs: Clerk security
• Stripe
  Role: Processes payments
  Receives: Your card details, entered on Stripe-hosted pages. Card numbers never touch our servers; we keep only a payment reference and the amount.
  Docs: Stripe security
• OpenAI
  Role: Reads the document text (the AI step)
  Receives: The text of your document, to identify the allegations. Under the OpenAI API policy, this data is not used to train their models, and their abuse-monitoring copies are deleted within 30 days.
  Docs: OpenAI API data usage
• Google (Gemini)
  Role: Powers the optional question-answering assistants
  Receives: The question you type and, for the report assistant, the contents of your report. Sent only when you use those features. We use the paid tier of the Gemini API, whose terms state that prompts and responses are not used to improve Google products or train models.
  Docs: Gemini API terms
• Resend
  Role: Sends our emails
  Receives: Your email address and the message text. Result emails contain a link to your results, never the contents of your document or report.
  Docs: Resend security
• PostHog
  Role: Measures how the site is used, and whether it is working
  Receives: Which pages you open and which features you use, a random visitor id, and your account id when you are signed in. Also: crashes and error reports, where people click and scroll, and how quickly pages load. For our AI assistants it receives only what a request cost us and whether it succeeded, never the question or the answer. It records how pages are used on screen, with every form field masked before it leaves your browser, and with recording switched off entirely on the upload, report, answers, saved-matters, account, admin, contact, and sign-in pages. It never receives your name, your email address, your search terms, your questions, or the contents of any document.
  Docs: PostHog security
• Google Ads
  Role: Measures which ads bring people to the site
  Receives: That a visit or a purchase happened, and the amount paid. It is loaded on public pages so we can tell which advertising works. It never receives your name, your email address, or anything about your document.
  Docs: Google advertising policy

What we do not do

• No selling or renting of your data, ever.
• No advertising or analytics service ever receives your uploaded document, the text taken from it, your search terms, or the questions you ask our assistants.
• No training of AI models on your uploads.
• No adding your document to the public collection, and no sharing it with other users, for any reason.

Payments

Payments are processed by Stripe on Stripe-hosted checkout pages. Card numbers go directly to Stripe and never touch our servers; we store only a payment reference, the product, and the amount.

Honest limits

No online service can promise zero risk, and we make no certification claims we have not earned. What we can say is specific: where your document lives, how it is encrypted, who can see it, when it is deleted, and which providers touch it, all of which is described above and kept current.

Contact & changes

For questions, or to remove an account and all its data, use the contact form. If this policy changes, the date at the top changes with it, and material changes will be announced on the site.