Source permissions
Choose every system OrthancIQ can read: product, billing, CRM, support, documentation, exports, or approved data-agent sources.
Security & trust
Adaptive intelligence. Controlled agency. One governed system.
Atlas can move quickly without becoming a black box. Every data source, retained field, action permission, approval threshold, and escalation path is scoped, logged, and reviewable.
REVIEW
READY
READY
Controls mapped
Guardrailed by design control set
AES
256
256
Encryption
TLS + AES-256
NO
TRAINING
TRAINING
Model training
Off by default
Security posture
Source permissionsRequired
Field and retention policiesRequired
Tenant isolationScoped per tenant
Encryption in transitTLS
Encryption at restAES-256
Audit loggingLogged
Least-privilege accessRBAC + MFA
Training on your dataOff by default
Data deletion on requestSupported
SOC 2 status. OrthancIQ does not currently claim completed SOC 2 certification. Current controls, architecture documentation, and the security roadmap are available under NDA.
Permission gates
You control what Brain Engine can learn from—and what Atlas can do.
Data access, retained fields, model inputs, action permissions, approval thresholds, and escalation policies are explicit parts of the OrthancIQ control plane.
Decision integrity
Every Brain Engine decision and Atlas action carries its evidence.
Brain Engine decisions remain tied to the customer signals and model state that produced them. Atlas actions remain tied to the decision, permission, owner, and approval policy that authorized them.
- No fabricated customer history or hidden data expansion.
- Brain Engine owns diagnosis and prioritization; Atlas owns the operating loop.
- Brain Engine only reasons from authorized customer context.
- New tracking and sensitive actions can remain approval-gated.
- Current controls and the security roadmap are available for review under NDA.
Controls
Security controls are part of the product, not a slide afterthought.
Tenant isolation
Each customer’s data is logically isolated, with boundaries enforced at the data and application layers.
Encryption
Data is encrypted in transit with TLS and at rest with AES-256, including backups where applicable.
Audit logging
Access and key actions are logged with timestamps and actor identity for review.
Access controls
Role-based access with least-privilege defaults and MFA for internal access.
Data retention
Retention windows for raw inputs and derived artifacts are defined during onboarding.
Deletion on request
You can request deletion of your data and model artifacts. We confirm completion and timelines in writing.
Backups & recovery
Encrypted backups and recovery procedures protect against accidental data loss.
Incident response
A documented incident-response process defines severity, ownership, escalation, and customer notification.
Security review readiness
We support vendor reviews with documentation, questionnaires, and an NDA-backed controls overview.
Data lifecycle
Your data and tenant-specific intelligence stay inside your boundary.
Every stage is scoped, encrypted, and logged. Customer data powers your tenant-specific customer-state model and intervention memory; it is not used for shared-model training by default.
Ingest
over TLS
Isolate
per tenant
Encrypt
AES-256
Model
tenant-scoped
Retain
defined window
Delete
on request
No shared-model training by default
Your data builds your model. We do not use it to train shared or cross-customer models unless you explicitly opt in, in writing.
Brain Engine artifacts follow your tenant
The value map, signal catalog, Brain Engine outputs, and Atlas evidence mappings are tied to your tenant and removed with your data on deletion.
For your security team
Need to run vendor review before sending data?
We can support a serious review process with a controls overview, security questionnaire, subprocessor list, incident policy, and architecture summary under NDA.
Controls overview (Guardrailed by design)
Security questionnaire (CAIQ-style)
Subprocessor list
Architecture & data-flow summary
Incident-response policy
Run the OrthancIQ security review.
Review Brain Engine inputs, Atlas action policies, architecture, and current controls before connecting the first customer signal path.