AI is a tool, not the proof.

AI-assisted tools can be useful for exploring directions, accelerating repetitive technical work, testing alternatives, supporting research, drafting internal working material or helping build workflow features such as classification, chat or voice.

The presence of AI in the process is not, by itself, a reason to use it or a customer-facing value proposition.

Generated or assisted output is reviewed against the actual project brief, source material, brand rules, technical constraints and evidence before it is treated as production work.

Claims, client outcomes, metrics, endorsements, project history and other factual proof must come from supported sources rather than generated inference.

Generated imagery or fictional concept work must not masquerade as client work, real staff, real customers, real campaigns or documentary evidence.

When generated concept media is useful in Graphaic Lab or exploratory work, it should be labelled according to its provenance and kept distinct from verified portfolio evidence.

Sensitive, confidential or regulated material requires project-specific handling rather than a blanket assumption that it can be placed into any AI service. Tool choice, data handling and access should match the requirements of the engagement.

If a production feature itself uses AI, the implementation should define its purpose, limits, failure states, human handoff and relevant data/consent constraints before launch.