DEJAN optimizes citations and grounding extracts through Bayesian optimization loops that iteratively test and score text edits against an active Rank Factor taxonomy using query potential scores to maximize visibility in AI answer engines.