Security
How we keep your workspace safe — and what we run it on.
Security is built into every layer of HeyAyra.AI. We follow modern secure-development practices and continuously review our infrastructure to protect your data.
Architecture
Every model call in the product passes through a single LLM gateway — self-hostable, auditable, and the only path to a model. Meetings run on a self-hosted LiveKit server. Your data lives in one Postgres cluster with row-level tenant isolation.
Encryption
All data is encrypted in transit using TLS 1.2 or higher. Sensitive data at rest is encrypted using industry-standard algorithms and managed encryption keys.
Authentication
Sign-in runs through Keycloak (OIDC, PKCE) and issues HttpOnly session cookies — tokens never touch browser storage. Sensitive actions like signing a document require a one-time passcode. Passwords are never stored in plain text.
Access Controls
Role-based access controls ensure users can only access the resources they are authorized to see. Four roles — super, system, site, end — gate every API at the edge. Administrative actions are logged and auditable.
Infrastructure
Our infrastructure runs on hardened cloud environments with automated patching, network segmentation, and DDoS protection.
Monitoring and Incident Response
We monitor for suspicious activity around the clock and maintain an incident-response plan to quickly investigate and remediate any security issues.
Reporting Issues
If you discover a security vulnerability, please report it responsibly to security@heyayra.ai. We investigate all reports promptly.
Questions? security@heyayra.ai · Request an invite →