Security at Evidaro

Built for work that must stand up to scrutiny.
Here is exactly what we do.

This page lists controls that are implemented in the product today – and, just as clearly, what we don’t claim.

Implemented controls

Security in the architecture, not the brochure.

Tenant isolation

Every organisation’s data is protected by PostgreSQL row-level security, in addition to permission checks in every service.

Role-based access

Seven roles with company scopes, enforced in APIs, search, the assistant, exports, notifications and background jobs.

Separation of duties

Preparers can’t approve their own work. The assistant and automated jobs can never approve, close or sign.

Sessions and passwords

Passwords hashed with scrypt; random session tokens stored only as hashes in HttpOnly cookies.

Protected documents

Originals stored write-once with SHA-256 hashes, downloaded only as attachments; retention and legal hold supported.

Untrusted content

Uploaded content is treated as evidence, never as instructions. Embedded instructions are flagged.

Safe retrieval

Only public web addresses are fetched; private networks and cloud metadata endpoints are blocked.

Tamper-evident trail

Every significant action is hash-chained per organisation and can be verified by administrators.

Processing transparency

Administrators see where files are stored, whether OCR runs and which AI provider, if any, is used.
What we do not claim

Honest about the limits.

Trust starts with not overstating. To report a security issue, write to security@evidaro.ai.

No certifications yetEvidaro holds no third-party security certification at present.
Data locationWe do not claim India-only processing: data location depends on your deployment and any AI provider you enable – all shown in Settings.
Encryption at restDepends on the configured storage and database.
MFA and SSONot yet available.

See the controls in the product.

A private demo workspace with fictional companies. No sign-up.

Try the live demo →