Trust and control
How NYEDZ uses artificial intelligence
This page distinguishes deterministic rules, simulated features and future external models. It describes the data used, limitations and human control.
Information reviewed on August 16, 2026.
Features in scope
Inventory of features presented as intelligent or automated.
Local ranking of situations such as urgent tasks, overdue invoices, quotes to follow up, new leads and upcoming appointments.
Data used: aggregate counts and business references in the active workspace.NYEDZ prepares one precise internal action. A sensitive action remains blocked until an authorised person approves it.
Data used: business target, age, status and configured threshold.Voice is used only to recognise a supported navigation destination. It does not approve payments or business writes.
Data used: navigation command spoken in the browser.Technical safeguards
These protections are enforced by the application and verified by tests.
- Every automated output is structured and validated before use.
- A model can never modify a business table directly.
- A sensitive action requires explicit human approval.
- Relevant proposals, decisions and executions are audited without logging secrets.
- Production configuration rejects any external AI provider that is not implemented and audited.
Your controls
A recommendation remains decision support and can be challenged.
- Review the recommendation, rationale, data used and confidence level.
- Reject an approval request without executing the proposed action.
- Review the workspace audit log when your role permits it.
- Report an incorrect recommendation through the contractual support channel once approved.
This information is not a regulatory validation. Any future activation of an external model will require a new dated review and an update to this page.