Vicaro Trust Center

Security, data privacy, and governance for every AI employee.

Compliance,
at a glance.

Current status, including audits still in progress.

SOC 2 Type II

Audit in progress

Audit in progress with Scytale. No report has been issued yet.

ISO/IEC 27001

Audit in progress

Audit in progress with Scytale. No certificate has been issued yet.

GDPR

Compliant

Our Privacy Policy explains lawful bases, data rights, and subprocessors.

CCPA

Compliant

Our Privacy Policy explains California privacy rights. We do not sell personal information.

Security &
architecture.

Where data lives, who processes it, and how AI employees are governed.

Data protection

  • Hosted in the EU

    Compute, storage, and the database run in Frankfurt (eu-central-1).

  • Encrypted

    Keys, secrets, and integration credentials are encrypted at rest with AES-256. All data in transit uses TLS.

Architecture

  • Subprocessors

    Ask our security team for the services that help us deliver Vicaro, where they process data, and which optional providers you can enable.

  • AI security controls

    Policies, human approvals, separate identities, and an audit trail govern what AI employees can do.

Security in four layers

Execution policies

  • Governance policies

    Define which actions agents may take. Allow, warn, require human approval, or block an action before it runs.

  • Guardrails

    Baseline protections govern tool connections, integration access, and policy changes. Add presets that fit your organization.

  • Prompt-injection protection

    Scan untrusted messages and tool results. Feed risk signals into policies to block suspicious content or require approval.

  • Kill switch

    Say “stop” to halt a running task. Pause an agent to prevent new messages and scheduled work from starting.

Identity & access

  • Accountable human

    Every agent has a named manager who sets direction and can approve or deny sensitive actions.

  • Separate agent identities

    Provision separate identities and credentials for agents. Keep their access distinct from human accounts.

  • Communication controls

    Choose who can reach each agent, internally and externally. Restrict access to specific people and channels.

  • Org-wide controls

    Apply policies across every agent. Organization-level restrictions cannot be loosened by an individual agent’s rules.

Data & providers

  • Full auditability

    Review actions, policy decisions, and configuration changes in a tamper-evident audit log. Export records for independent review.

  • No training by default. You own your data.

    Training-data collection is off by default. Your organization controls whether to explicitly opt in.

  • Full provider control

    Choose approved model vendors, pin models per capability, and disable providers your organization does not allow.

  • Data-flow transparency

    Inspect provider and model choices, review subprocessors, and restrict processing to providers you approve.

Infrastructure & limits

  • Budgets & cost controls

    Set spend caps per agent and scheduled task. Track usage and stop work when a limit is reached.

  • Bring your own keys

    Use your own model-provider credentials so inference runs under your account and commercial terms.

  • Dedicated tenants

    Choose a dedicated data plane with isolated storage, secrets, and independently revocable tenant access.

  • Multi-cloud & BYOC deployment

    Deploy a dedicated data plane in your own AWS or Google Cloud account and chosen region, with private networking and encryption keys you control.

Documents for
each stage.

Public policies are here to read. Ask our security team for agreements.

Talk to our
security team.

Questionnaires, audits, and vulnerability reports.

Security reviews

Send your questionnaire or audit request. Implementation detail is shared under NDA.

Report a vulnerability

Email the details and steps to reproduce. Please give us time to fix it before sharing it.

What would you hand over first?

Start with one role and one real task.

Apply for access

Takes about a minute. We reply by email.

What’s your name?

Press Enter to continue