This experience paper shows how the AI-teach consortium translated a 'Mapping Insights' study of Artificial Intelligence in Education (AIED) into actionable content portfolios for capacity building in teacher education. The mapping was based on four sources: (1) a literature review; (2) an analysis of 21 European best practices; (3) curriculum screenings of eleven teacher-education programmes across six countries; and (4) focus group interviews with 48 stakeholders (teachers, teacher educators, policy actors and ICT roles). Across these sources, findings showed some recurring gaps: fragmented and elective-only curriculum coverage; limited subject-specific pedagogical integration; underdeveloped learning analytics competency; unclear assessment standards for AI-supported work; optional and inconsistent professional development; and uneven governance, infrastructure, sustainability and equity measures. The gap matrix consolidates these insights into nine dimensions and strategic recommendations. These are used to develop three modular content portfolios aligned with Holmes' three domains (learn about AI; learn with AI; use AI to learn about learning), and linked to DigCompEdu and UNESCO's AI Competency Framework for Teachers. We conclude with implementation implications for institutions seeking coherent, ethically grounded and scalable AI integration.