The EU AI Act's risk-classification rules apply just as much to the AI quietly running in your back office as they do to a customer-facing feature. This keynote opens day two with what that actually means for product teams. The talk covers how to classify AI features correctly under the Act, and how to embed intelligence into legacy and vertical software without triggering a compliance review you didn't plan for. Attendee takeaways - How to classify back-office AI features under the EU AI Act's risk tiers - Common compliance triggers teams don't see coming - A checklist for embedding AI into legacy software without a green-field roadmap