Checking agent
The checking agent is a separate AI agent that judges tone and brand fit.
Code checks what has one right answer, such as a character limit. The checking agent judges tone and brand fit, and it also flags any other sentence that reads as a claim.
In one engineer's experiments at Anthropic, agents reliably rate their own work too generously on judgment calls such as design, and a separate checking agent was far easier to make skeptical. So the agent that drafted the work never grades it alone.
Before you trust it, run it on past drafts your reviewers approved and rejected, and correct its instructions until it agrees with them on both. Where you can, run it on a model from another family that is at least as capable as the drafting model. In one study on problems with right answers, such as math, a model checking work from its own family was biased toward accepting wrong answers.
Example
In a paid social example, code catches a headline over the character limit and an offer that isn't among the client's approved offers. The checking agent flags one brief's tone and one claim the drafting agent worded its own way. The drafting agent fixes the headline, the offer, and the tone. Legal then approves new wording for the flagged claim.