Privacy notice

Draft for pre-launch review. Operator organization: [TO COMPLETE]. Privacy and security contact: [TO COMPLETE BEFORE PUBLIC LAUNCH]. This is not a finalized legal notice.

What this service processes

Supabase processes account sign-in and session information. The application stores a local identity, account plan/status, and operational usage records. Research questions, paper metadata, abstracts and uploaded document content may be processed to provide research assistance. Supplied analysis content may be sent to the configured AI provider; requests ask the provider not to store model responses, but this is not a promise of zero provider retention.

Saved work and temporary documents

Projects, thesis plans, selected paper metadata, derived evidence and reports are saved when you use their save or planning actions. Evidence Analyser V1 stores uploaded PDFs and extracted text in your selected project library (or an automatically created Evidence library project). Completed per-paper analyses and two-paper comparisons are saved automatically for reuse. Removing a library PDF deletes its linked V1 analyses and comparisons; deleting its project removes the library. Extraction uses temporary files, normally removed when processing finishes, including handled failures. Abrupt worker termination can delay cleanup until ephemeral storage is removed.

Essential storage and operations

The browser uses essential Supabase session storage and application preferences/state. No tracking analytics or advertising cookies have been added. Approximate daily signed-in-user counts use date-scoped hashed internal identifiers in short-lived operational Redis state; no new browser tracking identifier is introduced. Operational logs and model-use accounting record categories, timing, counts and token usage rather than research content. Literature caches contain public bibliographic information. The full-text resolver also caches machine-accessible article text and source/access metadata for up to 24 hours, and unsuccessful lookups for up to five minutes. User-uploaded documents are excluded from this shared resolver cache. Hosting, authentication and AI providers have their own processing/retention terms.

Deletion and retention

You can delete owned projects and planning content using the application controls. Account deletion currently requires an operator request through the contact above, once completed. Local deletion removes linked projects, analyses, plans, reports and user usage records. A hashed identity marker prevents accidental reprovisioning; privileged audit records and non-user-linked aggregate model accounting remain. External Supabase identity deletion is a separate administrator step and will be reported as pending until completed. Backup copies age out under the operator's configured backup policy; deletion is not a promise of instant erasure from backups.

Before launch, the operator must publish applicable retention periods, contact details, provider disclosures and jurisdiction-specific rights/processes. Do not upload patient-identifiable or other sensitive personal information without a separately approved processing arrangement.