Security architecture

Security is the foundational question enterprise IT and risk teams ask when evaluating automated AI. Rival Enterprise protects customer data, user identities, and system executions through an infrastructure-level security architecture paired with granular administrative controls.

Verify before deployment

This page outlines the technical security framework for Rival Enterprise. Specific encryption standards, certifications, and infrastructure parameters should be reviewed and verified by your internal security and compliance team prior to final deployment.

Data protection and encryption

Rival protects enterprise data across all lifecycle states using industry-standard cryptography and network isolation.

• Data in transit — All traffic moving between client applications, external APIs, and Rival control planes is encrypted using TLS 1.2+ over HTTPS.
• Data at rest — All stored databases, execution state logs, and vector storage namespaces are encrypted using AES-256 encryption.
• Zero model training on customer data — Customer inputs, agent prompts, workspace data, and output payloads are strictly isolated and are never used to train public foundation models or third-party AI systems.
• Logical and network tenant isolation — Customer data is logically segmented by tenant identifiers, ensuring strict boundaries across multi-tenant cloud environments.

Identity and authentication controls

Platform security begins with strict identity verification before any user or agent executes an action.

• Managed sign-in — Users sign in through Rival’s authentication service. Single sign-on through your own identity provider (SAML / OIDC) is not available yet.
• Immediate access removal — An admin can suspend a member or remove them from the workspace; a suspended member can no longer run anything. See User lifecycle.
• Token-based connector authentication — Connectors authenticate to external business systems via scoped API tokens or OAuth 2.0 grants, never by storing raw user passwords.

The shared-responsibility model

Enterprise security operates on a shared-responsibility framework: Rival secures the platform infrastructure, while your organization governs workspace configuration and usage.

Your responsibility

• User role and access assignments
• Connector approvals and least-privilege
• Global guardrail and tool policies
• Audit log monitoring and reviews

Rival's responsibility

• Platform infrastructure security
• Encryption in transit and at rest
• Tenant isolation and code sandboxing
• Authentication and session security

Security controls matrix

Security layer	Technical implementation	Enterprise risk mitigated
Data encryption	TLS 1.2+ in transit; AES-256 at rest	Prevents unauthorized data interception and eavesdropping
Access control	Role-based access control (RBAC)	Prevents unauthorized privilege escalation across tools
Action safeguards	Agent needs-approval guardrails and admin access requests	Intercepts high-consequence operations before execution
System visibility	Time-stamped, immutable audit logs	Guarantees compliance traceability and incident response

A quick example: mapping a security review

Consider an enterprise IT security team conducting a vendor risk assessment.

1. Checklist mapping — The security team maps their compliance checklist against Rival’s security architecture, verifying encryption standards and tenant isolation.
2. Policy verification — The team confirms that administrators can enforce least-privilege roles, block tools and regions they have not approved, and require human sign-off on the agent actions that matter.
3. Audit readiness — The review concludes with clear evidence that every system action generates a verifiable audit log entry, turning vendor risk evaluation into a straightforward verification exercise.

Next steps

AI security and risk management The threats specific to AI systems, and how the platform mitigates them.

Approval workflows Review the human sign-off checkpoints that gate high-consequence actions.

Guardrails and allowed models Configure global model limits and data privacy rules.

Audit logs Review platform activity tracking and the immutable execution record.