Security
Rubin is built for teams that need clear authorization boundaries, reviewable operations, and careful handling of company data. We design the platform around encryption, workspace-scoped access, auditable activity, and disciplined release processes.
Platform controls
Core platform controls are designed to keep data encrypted, access bounded, and workspace operations reviewable.
All data is encrypted in transit with TLS 1.3 and at rest with AES-256. Database fields that contain credentials or tokens are additionally encrypted at the application layer before they reach storage. Backups follow the same envelope-encryption scheme.
Rubin stores the entities, relationships, and operational data you bring into the platform — blueprints, registries, uploaded files, conversation history, and agent execution logs. We do not store raw model weights or training data from your prompts. File attachments are kept in isolated object storage keyed to your workspace.
Passwords are hashed with bcrypt (cost factor 12). Sessions use signed, HTTP-only, secure cookies with short TTLs and automatic rotation. Role-based access control (RBAC) scopes every API call to the permissions granted by your workspace admin. SSO via SAML 2.0 and OIDC is available on enterprise plans.
Each workspace operates in a logically isolated tenant. Queries are scoped at the database layer so one workspace can never read another's data. Agent executions run in sandboxed environments with no cross-tenant network access.
You can delete individual records, entire registries, or your full workspace at any time from the settings panel. Deletion is propagated to all replicas and backups within 30 days. On request, we issue a signed confirmation once purging is complete. You can also export all your data in standard formats before or instead of deletion.
When you use AI features, prompts and context are sent only to the model providers you enable for your workspace. Rubin does not use your prompts or documents to train public foundation models, and we do not sell customer data. Provider-specific terms apply to inference; we document subprocessors and data flows in our Data Processing Agreement (DPA) and can share a current list under NDA for procurement reviews.
Operating principles
We aim for strong defaults, explicit authorization, and operational transparency rather than marketing-only assurances.
Workspace permissions are meant to expose only the data and actions each user or system needs.
Important actions are logged so teams can trace what changed, who initiated it, and how the system responded.
Deletion, export, and security review workflows are available so customers can manage data on their own terms.
Operational practice
Security also depends on deployment discipline, vendor management, and clear response processes when events need attention.
Every mutation — entity creation, permission change, agent execution — is recorded in an immutable audit log scoped to the workspace. Logs are retained for the lifetime of the workspace and can be exported for internal reviews, investigations, and customer security discussions.
Rubin runs on hardened cloud infrastructure with automated patching, network-level segmentation, and DDoS mitigation. Production systems are monitored continuously with anomaly detection and on-call response. Changes to production systems pass through peer review, automated checks, dependency scanning, and staged rollouts.
Production workloads run on major cloud regions with redundancy and encrypted volumes. Enterprise customers can align deployment choices with residency and data-processing requirements where technically feasible. We maintain an inventory of subprocessors (hosting, email, analytics where used, model APIs) and notify material changes in line with our agreements.
We maintain incident response runbooks, on-call rotations, and post-incident reviews. If an incident may affect your workspace, we notify you without undue delay with what we know, what we are doing, and what you should do next. Enterprise agreements can define contact paths, severity thresholds, and joint tabletop exercises.
We support vendor security questionnaires, architecture reviews, and shared-responsibility discussions with your security team. We can share current documentation about our controls, data flows, and subprocessors as appropriate for procurement and technical review.
If you discover a security issue, please reach us at security@rubin.zone . We triage reports within 24 hours and will keep you informed throughout the resolution process.
