amaise classifies all data into three tiers with progressively stricter protection measures. Each tier has a code (C1, C2, C3) used in our architecture diagrams and a semantic name used in policy and customer communication. Both refer to the same tier.

C3 — Sensitive (Strictly Confidential) — your uploaded documents and all the persistent content derived from them: OCR-processed PDFs, thumbnails, extracted full text, AI summaries, structured extractions, diagnoses, AI answers, chat, and the case notebook, together with the full-text search index and backups built from them. Your documents and their file-storage artifacts are encrypted under a dedicated per-tenant key, which enables crypto-shredding: disable the key and that tenant's documents become permanently unrecoverable. All C3 data is strictly tenant-isolated and encrypted at rest with AES-256.

C2 — Confidential — the case and document inventory (which cases and documents exist), internal identifiers and mappings, the audit trail, integration-agent events, and short-lived derived data (AI answering context and transient exports). This tier also covers your access credentials and SSO mappings (managed by our identity provider) and our infrastructure secrets. Stored encrypted at rest, access-controlled, never logged in plaintext, and rotated on a scheduled basis.

C1 — Operational (Internal) — Usage analytics, LLM token metrics, error reports, application logs, operational cache, and availability data. These contain no customer document content, are cleansed of personal data, and are subject to defined retention periods.

Input data and derived artifacts inherit the tier of their source — for example, OCR output derived from a customer PDF is treated as C3 / Sensitive. Tier assignments and storage mappings are documented in our internal Security Architecture and reviewed during the ISO 27001 / SOC 2 audit cycle.