Protection that lives inside the decision, not around it.
For a large analytical platform, information security cannot exist solely as an external control system. It has to be built into the mechanisms of data processing and decision-making themselves. Mantikon Guard manages access at the level of users, agents, objects, attributes and operations.
- Next waveDesigned as an application area; a demonstrated world model in this domain comes first.
- Who decidesDPO, CISO, CIO, legal
- Typical sourcesData catalogue, classification, access logs, processing purposes
- 01
Least privilege per agent
A planning agent may see staff availability without medical data; a financial agent may use aggregated costs without personal account detail.
Access - 02
Purpose and provenance
The platform records where information came from, what it may be used for and under what conditions, and keeps that with the decision.
Provenance - 03
Lifecycle and correction
Retention periods, detection of unauthorized access and the procedures for correcting or deleting data are part of the system, not a separate project.
Lifecycle
A worked
example
Illustrative scenario, not a customer case.
A council is convened on a staffing question. The agent that plans the roster receives availability and qualification data; it never receives medical records, and it cannot request them.
The financial agent in the same case works with aggregated departmental costs. Neither agent can widen its own scope, because the policy is enforced outside the models.
The decision that results carries the list of sources it used, the purpose under which each was read and the checks that were performed.
Scope and
measures
Example workflows
Dataset registry · purpose-based access · access review · retention and correction.
Evaluation metrics
Sensitivity classification coverage · unauthorized access attempts · time to fulfil a correction request · share of decisions with complete provenance.
Discuss data governance requirements