Purpose first
Any proposed AI use should serve a defined business purpose, not be added for its own sake. Scope and data use are agreed for each engagement.
Responsible AI
People remain responsible.
These principles guide how AI may be considered in a project. Specific controls and responsibilities are agreed in writing, not assumed from this page.
Human review
People should review consequential outputs and exceptions. The client retains authority over consequential business decisions.
Information boundaries
Client information should be used only for an agreed purpose and kept separate between engagements. Specific handling arrangements must be confirmed in the engagement agreement; this page does not assert a verified technical isolation guarantee.
Verification
Automated output can be inaccurate or incomplete. It needs checking against the relevant source and context; error-free output is not guaranteed.
External communication
Messages sent externally should have agreed boundaries and appropriate human oversight. No public claim is made here that all such communication is already automated or deployed.
