Read within the role
Agents should use only the programme information and tools their role is permitted to access.
AI AGENTS
Intelligence grounded in the programme, scoped to the role and accountable to the people doing the work.
AI AGENTS
Prepare updates, review submissions and develop creative directions with agents that work within each person’s permissions.
The agent works through three steps:
Sending this update requires your approval.
Agents should use only the programme information and tools their role is permitted to access.
Draft outputs and proposed actions should remain visible, with evidence and uncertainty available for human assessment.
Publishing, financial commitments and sensitive claims require explicit review. Agents support judgement rather than replacing it.