This paper presents Identity Before Tools as a full pedagogical framework for human-centred AI education. The central principle — that every introduction of AI must begin with the prior question of who this person is, what they value, and what they want to protect — is grounded in the Mirror Hypothesis (Working Paper 01) and tested across three case studies: MET Film School Berlin (BIMM University), Haus für Musik Potsdam, and SOS Kinderdorf Berlin (2024–2026). The paper presents five structural phases of the framework, documents the transfer effect observed across workshop settings, and develops implications for EU curriculum development and AI governance. Identity Before Tools is proposed as a structural prerequisite for safe and generative human-AI collaboration — measurable, teachable, and transferable.