One of the key pillars of intelligence is the ability to adapt to change. Intended for graduate students, researchers and practitioners, this book develops a principled framework for designing lifelong AI agents that can incrementally and efficiently acquire new knowledge and skills over time. It is organised around four main parts. First, it introduces continual learning agents, clarifying how learning systems interact with dynamic environments beyond the static training paradigm. Second, it develops the foundations of adaptation through online learning, transfer learning, pretraining, and self-supervised learning. Third, it focuses on memory consolidation and the stability–plasticity trade-off, presenting key algorithmic families-regularisation, replay and architectural methods-alongside multi-task and meta-learning strategies. Finally, it connects these foundations to practical recommendations, applications, and open challenges, including distributed continual learning, reasoning, foundation models, and never-ending AI learning agents.