Your filing is your filing.
You're considering uploading a draft brief to our servers. Here is exactly what happens to it, in plain language.
Plainly, step by step.
Where your document goes
Your filing is uploaded over an encrypted connection (TLS) and stored encrypted at rest, in AWS (US East, Virginia). Text is extracted on our servers; scanned pages are processed by our own OCR service, not a third party.
Who reads it
Software, not staff. No human at Fusion Collective reads your filing in the normal course of operation. Excerpts of your filing — the passages surrounding each citation — are sent to the model providers on our panel (Anthropic, OpenAI, Google) to perform the analysis, under API terms in which providers do not use the data to train their models.
What we keep, and for how long
The analysis and report are retained in your account so you can return to them. You can delete a filing and its analysis at any time from your dashboard. Citation verification results are cached — but the cache stores citations and their verification status, not your document.
Training
We do not train models on your filings. Ever.
Privilege and work product.
We designed Sentinel Citation so you can analyze drafts before filing. We are not your law firm, and uploading does not create an attorney-client relationship with us; consult your own confidentiality obligations. What we can tell you: your document is not shared with anyone outside the processing chain described above, and your report is visible only to your account.
Encrypted in, encrypted at rest, read by software, never used for training.
If your matter requires more — retention agreements, a security questionnaire — write to us: info@fusioncollective.net