Separated client spaces
Data is organised by space and scope, with clear separation between client contexts, administration and operational work.
Security & data
Torus treats client documents, exchanges and evidence as sensitive material. The platform favours separation, minimisation, human validation and no default reuse.
Data is organised by space and scope, with clear separation between client contexts, administration and operational work.
Client documents are used by the assistant only when they are enabled for the relevant scope.
Torus can help review uploaded evidence and flag what appears sufficient, incomplete or inconsistent; validation remains human.
Reports, workbooks and evidence are designed to be reviewed, shared and retained as working material.
Principles
Security is not decorative copy: it shapes how the assistant, e-learning, risk analysis and evidence workflows are designed.
Data is organised by space and scope, with clear separation between client contexts, administration and operational work.
Client documents are used by the assistant only when they are enabled for the relevant scope.
Torus can help review uploaded evidence and flag what appears sufficient, incomplete or inconsistent; validation remains human.
Reports, workbooks and evidence are designed to be reviewed, shared and retained as working material.
AI & confidentiality
AI exchanges must remain aligned with the expectations of a CISO, cyber consultant or compliance team.
ZDR means Zero Data Retention: client data sent to the model during AI exchanges is not retained for training or reuse.
By default, Torus does not reuse client documents, evidence or conversations to train models or enrich datasets.
Any potential contribution to a Torus reference set would need a separate, explicit, minimised, redacted and validated contractual process.
When client documents are enabled, they remain the primary source; Torus cyber and compliance references support without replacing client context.
Clear boundaries
Torus does not promise automatic compliance.
An AI answer does not become evidence by itself.
Removing a name from a document is not enough to call it anonymous.
Generated deliverables remain drafts or working material to validate.
Risk, audit and compliance decisions remain under human responsibility.
A useful demo can also cover your data, validation, evidence and scope-separation constraints.