Governance

Autonomous does not mean unsupervised. Agents execute; humans authorise. No campaign phase advances to the next, and no asset is exported for activation, without an explicit human approval recorded against a named user.

• 01Strategy gate — positioning, audience and KPI framework must be approved before creative production starts.
• 02Creative gate — every asset set is reviewed against brand guidelines before it enters the media plan.
• 03Media gate — allocation, flighting and budget must be approved before any activation package is produced.
• 04Change gate — material changes to an approved campaign reopen the corresponding gate.

Access is granted by role, not by seniority. Strategists, creatives, media planners, brand owners and executives each receive the minimum visibility their function requires. Approval authority is assigned per brand and cannot be self-granted by an operator.

Every agent run is recorded: the inputs, the retrieved sources, the model decisions, the output version and the human who approved or rejected it. Audit records are immutable and retained for the life of the account plus the retention period defined in the Privacy Policy.

All generated output passes a brand safety layer validated against the client’s own guidelines and against category restrictions. Regulated categories — including financial services, health, alcohol, gambling and content directed at children — require additional human review before any output leaves the platform.

Any user may halt an agent run at any point. Suspected brand-safety failures, factual errors or governance breaches are escalated to the platform governance owner, investigated, and reported back to the affected client with the corrective action taken.

Widemedia Group AI Ltda. remains accountable for the operation of the platform. Responsibility for the commercial decision to publish an approved campaign remains with the client and the named approver.