RAMUS™ Executive operates on a zero-retention architecture. Documents you upload are processed in volatile server memory. Your Decision Pack is generated in a temporary workspace that exists only for the duration of your active session. The moment you download your files, close the browser, or your session ends — all data is permanently destroyed.
There is no assessment history, no stored copies, and no ability for anyone — including RAMUS staff — to retrieve your documents or analysis after the session ends.
Documents are processed via the Anthropic API under their commercial terms. Anthropic does not store or train on data submitted through their API.
We log three data points per assessment for billing purposes only: the access code used, the timestamp, and whether the assessment completed successfully. We do not log what you analysed, who the parties were, what the findings were, or any content from your documents.
RAMUS's proprietary analysis methodology — the Strategic Challenge Layer, Creative Pivot™, and Decision Brief generation logic — is injected server-side. Analysis prompts never reach your browser. Your browser communicates with our server; our server communicates with the AI. The intellectual property that powers RAMUS is protected at the infrastructure layer.
| Sub-Processor | Purpose | Data Access |
|---|---|---|
| Anthropic PBC | AI analysis (Claude) | Document text during processing only |
| JSONBin.io | Access code management | Codes and usage counts only — no documents |
| Replit / Hosting Provider | Server infrastructure | Volatile memory during session |
RAMUS is built to meet the standards of the Legal AI Evaluation Framework published by Legal Benchmarks (legalbenchmarks.ai). Our self-assessment scorecard is available on request.
RAMUS does not create or store vector embeddings of your documents. No human at RAMUS reviews your documents or analysis outputs.
Security concerns: security@ramus.co.za
General enquiries: info@ramus.co.za