Skip to main content

Security Architecture & Zero-Trust Model

The Zero-Trust Philosophy​

Zero-trust in software security is a security model where no system component is automatically trusted, even if it is inside the application's own infrastructure. Every request, user, service, and process must continuously prove that it is authorized before gaining access.

Traditional systems use a perimeter trust model. Once something is “inside” the network or backend, it is often implicitly trusted. Zero-trust removes that assumption.

Implementation in Case Digestor​

In practical terms for software systems:

  • A user is not trusted simply because they logged in once. Every request must carry valid authentication and authorization context.
  • An application server is not trusted simply because it is part of the backend. It must authenticate itself before accessing the database or external services.
  • A database does not trust queries by default. It enforces Row Level Security (RLS) or equivalent policy checks on every query.
  • External services (such as AI APIs) receive only explicitly approved data, never full internal state.

If any layer is compromised, lateral movement is restricted because no internal component has blanket trust.

Applied Security Controls​

Applied to Case Digestor, zero-trust means:

  1. The browser never talks directly to the database.
  2. The server never queries data without passing RLS checks.
  3. AI services never receive raw confidential documents without filtering.
  4. Every API route validates session tokens on each request.

The benefit is containment. Even if an attacker gains partial access, they cannot automatically escalate across the system.

In short: assume breach, verify everything, restrict movement.

Privacy and Data Governance​

We understand that legal research often involves sensitive or proprietary strategies. Case Digestor is built on a "Privacy-First" architecture:

  • Data Isolation: Your personal folders, notes, and generated digests are protected by Row-Level Security (RLS). No other user—nor the administrative staff—can view your personal research without your explicit action (like sharing a link).
  • Technical Sovereignty: We believe you should always own your intellectual work. This is why we provide bulk export features (JSON) and PDF downloads. Your data is not "locked" within our platform.
  • AI Anonymization: Text sent for digestion is processed securely. Our AI models do not "learn" from your private notes or folder structures; improvements to the system are based on public jurisprudence datasets only.