No certification claim
The site describes intended controls and does not invent achieved certifications.
Sensitive by default; controlled by design
PDICON Intelligence security is designed around tenant isolation, role-based and document-level access, encryption, auditable human and model actions, explicit training consent and customer-defined deployment boundaries.
Direct answer
PDICON Intelligence security is designed around tenant isolation, role-based and document-level access, encryption, auditable human and model actions, explicit training consent and customer-defined deployment boundaries.
Design boundary
The site describes intended controls and does not invent achieved certifications.
Vector and graph systems inherit the same access rules as source documents.
Customer data contribution requires a recorded policy and purpose.
Every request resolves tenant, project, role and document permissions before data reaches retrieval or a model.
Input artifacts, model version, execution policy, output, expert decision and eventual outcome remain traceable.
Training eligibility, retention, region and self-hosted execution follow explicit policy rather than implicit reuse.
Operating matrix
| Subject | Input | Intelligence operation | Human / policy control | Output |
|---|---|---|---|---|
| Owner/Admin | Organization policy | Manage identity and controls | Administrative scope | Governed tenancy |
| Engineer | Assigned technical artifacts | Review and approve | Project/document scope | Recorded decision |
| AI operator | Approved dataset and model metadata | Train and deploy | Purpose-limited role | Controlled runtime |
Questions answered
No. Training contribution is controlled by explicit customer choice and policy.
No. Access is scoped by tenant, role, assigned project and document permission.