General-purpose AI can produce polished business-case prose in minutes. The difficulty is that fluent writing and defensible reasoning are not the same, and reviewers assess the reasoning.
A generic model does not inherently understand the Five Case Model or the way Green Book guidance treats appraisal, place and distributional impact. Structural problems can remain hidden until scrutiny begins.
The real bottleneck is often the gap between a plausible draft and what the framework requires. BCG embeds the methodology from the start so reviewers can focus on the quality of the case, not repair its structure.