Large language models (LLMs) have transformed learning and educational practices, yet concerns persist about whether authentic learning occurs when cognitive tasks are outsourced to artificial intelligence (AI) systems. We examined how AI systems can support educators in facilitating student’s authentic learning. This paper reports on the first two echelons of our echeloned design science research (eDSR). We evaluated 200 AI systems deployed across European educational institutions and interviewed 11 experienced educators in three countries. Based on the findings, we formulated and validated meta-requirements for AI systems that support authentic learning from the perspectives of students, educators, and educational institutions.