Home Features Integrations Security Log In Book a Demo
SECURITY TRUST CENTER

Security & Trust
Center

Transparency builds trust. Here's how we protect your data, your clients' data, and your compliance posture — at every layer of the stack.

AES-256
Encryption at Rest
TLS 1.3
Encryption in Transit
RLS
DB-Level Tenant Isolation
0
Data Sold or Shared

Multi-Tenant Security Architecture

Data isolation is enforced at the database level, not the application layer. Every query, every session, and every stored record is scoped to your organization.

Row Level Security

Every query is scoped to your organization via PostgreSQL RLS policies. Data isolation is enforced at the database level, not the application layer. Even if application logic had a bug, the database would refuse cross-tenant access.

Tenant Isolation

Each organization operates in a fully isolated context. Session-level org binding ensures zero cross-tenant data leakage. Switching organizations triggers a full session regeneration to prevent context bleed.

Encrypted Storage

All data encrypted at rest (AES-256) and in transit (TLS 1.3). Database connections use SSL with certificate validation. File storage uses Supabase's managed encrypted object storage.

Our Compliance Posture

We hold ourselves to the same standards we help you achieve.

Data Residency

Choose your data region. Supabase projects available in US, EU, and APAC. Data stays in your selected geography.

Encryption at Rest

AES-256 encryption for all stored data via Supabase's managed PostgreSQL. Application-level secrets use AES-256-GCM with unique IVs per record.

Encryption in Transit

TLS 1.3 for all API communications. HSTS enforced. Database connections require SSL with certificate validation — plaintext refused.

Access Controls

Role-based access across Admin, Manager, Analyst, and Viewer tiers. Principle of least privilege enforced. Permission checks fail-closed on null.

Audit Logging

Every sensitive operation logged with user identity, timestamp, and action detail. Logs are org-scoped and immutable via append-only patterns.

AI & LLM Security

Our AI engine is designed with security-first principles. Every input sanitized, every output validated, every key encrypted.

Prompt Injection Defense

All user and external content passes through a dedicated input sanitization pipeline before reaching AI models. The pipeline detects multiple categories of injection patterns — role hijacking, instruction override, system reveal, XML injection, and context overflow. Hostile patterns are replaced with [REDACTED], creating an audit trail.

Multiple injection patterns blocked Audit trail on detection

Input / Output Validation

AI-generated values (severity, categories, types) are validated against whitelists before database insertion. No raw AI output reaches storage. Messages and conversation length are capped to prevent abuse. Malformed JSON triggers validation errors, not pass-through.

Enum whitelisting 4K char cap Multi-stage JSON parser

Rate Limiting

Database-backed rate limiting on all AI endpoints — per-organization rate throttling prevents abuse. Per-user and per-org throttling is enforced at the database layer. Rate limit state is stored in PostgreSQL, not in-memory, so it survives server restarts and horizontal scaling.

DB-backed (not in-memory) Per-org throttling

BYOK Support

Bring Your Own Key — use your organization's API keys for AI providers. Keys are encrypted at rest using AES-256-GCM with unique initialization vectors per record. Keys are never returned in API responses, never logged, and never visible after storage.

AES-256-GCM encrypted Never returned in responses

Data Handling & Privacy

What we collect, how it's stored, how long we keep it, and what you can demand we delete.

What We Collect
Organization profile, user accounts (email, display name, role), compliance assessment data, uploaded evidence documents, and AI-generated analysis tied to your assessments. We collect the minimum necessary to deliver the service.
How It's Stored
Managed PostgreSQL with Row Level Security enabled on all tenant tables. pgvector for semantic embeddings, scoped to your org. Evidence files in encrypted object storage. All tables have dedicated tenant identifier columns — isolation is structural, not relational.
Retention
Data is retained for the duration of your active subscription. Upon account closure, all data is scheduled for full deletion within 30 days, including backups, vector embeddings, and file storage.
Right to Deletion
Request complete data removal at any time by contacting the security team. All data, pgvector embeddings, uploaded files, and backup snapshots are purged within the commitment window. We provide written confirmation on completion.
No Data Selling
We never sell, share, or monetize your data. Period. Your compliance posture is your competitive advantage — we have no interest in it beyond delivering the service.

Responsible AI Commitment

We believe AI in compliance must be held to a higher standard. These are the commitments we make — and enforce in code.

No Training on Your Data

Your compliance data is never used to train AI models. We use commercial API access with strict data processing agreements in place with Anthropic and OpenAI. Your responses stay yours.

Model Transparency

We document which AI models power each feature. We use leading commercial AI models with strict data processing agreements. Model versions are pinned for consistency across your assessment cycles.

Temperature Calibration

Each AI feature uses near-zero temperature settings optimized for accuracy over creativity. Compliance analysis runs near-deterministic — outputs you can rely on across audit cycles. Every endpoint is individually tuned for its task.

Infrastructure Stack

Purpose-built on managed, enterprise-grade infrastructure with no single points of failure.

Supabase
PostgreSQL 15+
pgvector
Semantic search
Node.js / Express
API layer
Connection Pooling
Connection pooling
Authentication Layer
Identity & sessions
Supabase Storage
Encrypted file storage

Questions About Our Security?

Our team is happy to discuss security practices, share compliance documentation, or arrange a security review call with your team.