“(unknown current hero — generic AI-tool pattern)”
Lawyers evaluating AI have one overriding concern: confidentiality. The hero doesn't address it. "Will my client data be used to train models?" is the question that kills 80% of legal AI evaluations. Waiting until the security page to answer it loses in-house counsel who never scroll that far.
“Add a trust line directly in the hero: "Your data never trains our models. SOC 2 Type II. Attorney-client privilege preserved."”
Pre-empts the #1 objection before it forms. Three claims (no training, SOC 2, privilege) map directly to the three concerns every GC raises in procurement review.