This vision paper outlines a future learning ecosystem in which AI agents act as tutorial companions, analyse learning progressions and make adaptive recommendations. The use of retrieval augmented generation (RAG) systems will make learning content accessible in a contextualised manner and generate individual feedback. Initial prototypes from the corporate world demonstrate the technical feasibility of this. We therefore assume that if AI-supported agents are successfully trialled, everyday teaching and learning at universities can be fundamentally changed in the direction of self-directed learning.