Security

Overview

Security in Aqili Next is structural. The same controls that protect your data also make the platform's AI safe to use. For a higher-level summary aimed at security reviewers, see the Security overview.

Tenant isolation

Aqili Next is multi-tenant with isolation enforced by Postgres row-level security. Every tenant-scoped query is constrained at the database, so one operator can never read or write another's data — even if application code had a bug. Agents inherit the same boundary.

Authentication and SSO

Operators sign in with email and password or through single sign-on (SAML/OIDC) tied to your identity provider, where you can enforce MFA and lifecycle policies. Parkers authenticate separately through the portal with email verification.

API keys and scopes

Integrations use API keys scoped to exactly the access they need, following least privilege. Keys can be rotated and revoked, and their use shows up in the audit log like any other actor.

Audit logging

Meaningful actions — by users and by agents — are recorded with who, what, and when. The log is queryable and exportable, giving you an accountable record for compliance and investigation.

AI safety

Every agent action that changes data is gated behind human approval and recorded for replay. Agents operate only within your tenant, and we configure AI providers to disallow training on your data. See AI agents for the full model.