Skip to content
AnvelitBUILD BEYOND
TRUST CENTER

Security, Audit

and Compliance

One bad autonomous action destroys credibility. Conservative defaults, human approval on high-severity work, and a permanent audit trail are non-negotiable.

Security

Conservative defaults, least privilege and encryption at rest and in transit across everything we build or operate.

RBAC

Role-based access on every system and scoped permissions on every automated action. Least privilege by default.

Encryption

Managed secrets, minimum IAM, encryption at rest and in transit on AWS infrastructure.

Audit logs

Append-only trails capture every decision and action. Readable, exportable, tied to what ran.

Human approvals

High-stakes automated actions stay behind a human gate. Autonomy is earned per action type.

Data isolation

Row-level tenancy. Customer data stays yours and is never used to train third-party models.

Governance

Permissions, approvals, audit trails and a written policy for what a system may do unsupervised.

Compliance roadmap

Privacy Act and GDPR posture in place. SOC 2 Type 1 in progress. Type 2 and SAML when production load justifies them.

Controls & compliance

Honest status chips - not every badge marked Operational.

Australian Privacy ActIn place
GDPR postureIn place
Role-Based Access ControlOperational
Row-Level TenancyOperational
Data IsolationOperational
Encryption at Rest & TransitOperational
Append-Only Audit TrailOperational
Human Approval GatesOperational
Data Never Trains Third-Party ModelsOperational
Governance ProgrammeIn place
AWS InfrastructureOperational
SOC 2 Type 1In progress
SOC 2 Type 2Planned
SAML SSOPlanned

Compliance sequencing

Before launch

Australian Privacy Act posture, GDPR posture, tenancy, encryption, audit trails.

~5 paying customers

SOC 2 Type 1 engagement begins - not before we have real production load.

Months 12-18

SOC 2 Type 2, SAML SSO and multi-region as enterprise demand lands.

Confidence-earned autonomy

Every decision returns structured JSON with confidence and reasoning. Autonomy is earned per action type. High-severity always escalates to a human. Prompts are regression-tested against 200+ frozen scenarios before any change ships.

Ask about security

Let's solve the operationalchallenges slowing you down.

Tell us where work stalls. The founding team replies within one business day.