English · Español

Security approach

Good automation includes clear limits, review and a way to recover. This page describes the principles we use to scope projects, not a certification or a universal service-level guarantee.

Separate clients, limited access

Each client and project is a separate boundary. We agree who may access which systems and data, limit access to the work, and define how it will be revoked. Credentials belong in an approved secret store, not source code, shared notes or public examples. Client production records are not used as public showcase data.

Review before action

We define validation rules, exception handling and the actions that need human approval. Material changes should be traceable without logging secrets or unnecessary personal data. Testing covers the agreed failure and duplicate-processing risks as well as the successful path.

AI with defined boundaries

Before AI processes client information, the project must define the permitted data, approved provider, retention settings, reviewer and fallback. Calculations requiring repeatability should use deterministic rules. AI output must be checked in proportion to its impact.

Support and recovery

Monitoring, notifications, recovery, manual override and shutdown procedures are selected for the workflow and documented in its scope. Support hours and response targets are agreed separately. At closeout, we agree on documentation, access revocation and the return or deletion of project data, subject to applicable retention duties.

Report a concern

Contact hello@wizautomations.net with a brief description. Do not include passwords, exploit payloads or client records in the first message. We will arrange an appropriate channel when more detail is needed.