Human approval is a primitive
SyncAI recommends. A named human accepts, rejects, escalates, or returns the recommendation. Direct plant execute is disabled. Approval, escalation, and accountability are product architecture — not a policy overlay.
Security & governance
Describe the controls that exist. Keep human approval in the architecture.
SyncAI is governed industrial intelligence. Security and deployment capabilities are described by implemented controls and validated configuration — not by a certification badge or a packaged on-prem SKU.
Implemented controls
Encryption, isolation, auditability, and named authority.
Audit log and decision traceability
Recommendations, evidence references, approval actions, and workflow state are designed to remain reviewable. The basis of a decision should not disappear into an opaque chat transcript.
Encryption
Operational information is encrypted in transit and at rest in the hosted environment. Customer-specific encryption and residency requirements are described against the actual engagement, not as a generic guarantee.
Row-level security
Multi-tenant isolation is enforced with row-level security (RLS) in the data layer, alongside application role boundaries. Customer configuration determines who can see and act on operational information.
Role-aware access
Access is scoped by role and the information boundary agreed for the workspace. Least privilege is the design intent. Enterprise SSO is a later product step, not a current capability.
Evidence before automation
The platform keeps observed evidence, assumptions, and missing information distinct. People accountable for the decision can see why a recommendation was made.
Human authority
The model is not the final approver.
Industrial recommendations can affect safety, production, cost, and asset life. SyncAI therefore treats recommendation and authority as separate layers. A named human remains accountable.
01
Recommendation
02
Evidence review
03
Named approval
04
Controlled work
Exact approval paths are configured for the customer’s operating model, roles, and risk boundaries. Direct plant execute stays off.
What we do not claim
Later product steps are not current capabilities.
SyncAI does not represent third-party certifications as complete unless they have been formally achieved and are current.
Enterprise SSO
Identity-provider SSO is a later product step. It is not a shipping capability on this site.
Packaged on-prem
There is no packaged on-premise product to order from this page. Deployment and data-handling requirements are defined during security and data discovery.
Third-party certifications
SOC 2, HIPAA, and similar attestations are represented publicly only when formally achieved and current. Architecture discussions distinguish implemented controls from certification status.
Data handling
Work from the customer-approved information boundary.
SyncAI is designed to use the records and systems the customer authorizes. Assessment work starts from customer-provided exports. Integration requirements for a later deployment are defined during security and data discovery — not assumed as live production connectors today.
We do not train foundation models on customer operational data without explicit consent. Competitive and operating information stays the customer’s.
Security contact
For vulnerability reports, architecture questions, or customer-specific control discussions against an agreed scope.