EU AI Act Art. 6 high-risk classification: how did your team document the borderline cases?

We're working through the Art. 6 classification for our AI systems and hitting the familiar grey areas: a recommendation engine that influences user choices but doesn't make final decisions, and a risk-scoring model that flags cases for human review. The regulation says high-risk applies when the system is a safety component of a regulated product OR used in specific domains (Annex III). Our use case sits in Annex III (employment, credit scoring adjacent) but the human-in-the-loop is designed to be meaningful — not a rubber stamp. How did your team document the borderline cases? - Did you classify as high-risk out of caution, or build a documented argument for limited-risk? - What evidence did auditors actually look at for the 'meaningful human review' claim? - Did the transparency obligations under Art. 50 change your documentation approach for the classification itself? Sharing peer experience here — this is about operational practice, not legal advice. What worked in your audit trail?