Learning to learn, a key transversal competence emphasized across AI literacy frameworks (e.g., DigComp 2.2, the AILit Framework, aiEDU, Digital Promise, and UNESCO’s AI Competency Framework for Students), fosters self-regulated and adaptive learning in dynamic environments. In agriculture, where climate change intensifies uncertainty, farmers must continuously adapt their practices by cultivating Self-Regulated Learning (SRL) skills. The ability to learn how to learn enables them to effectively monitor, evaluate, and refine their strategies, transforming experiential knowledge into sustainable action. This paper introduces a novel framework that integrates SRL with Large Language Models (LLMs) to support farmers’ adaptive learning. The framework is built on three core components: (i) Goal-Oriented Modeling: Explicitly represents farmers’ goals, risks, indicators, and contexts to facilitate structured practitioners’ learning experiences. (ii) Collective Knowledge Base: Enabling the reuse and sharing of farmers’ learning experiences to promote collaborative knowledge construction. (iii) LLM-Augmented SRL Feedback: Providing personalized, context-aware feedback across all SRL phases (e.g., goal setting, self-monitoring, self-evaluation, and metacognition), grounded in the Collective Knowledge Base, to enhance decision-making, reflective thinking, and autonomous learning. An application scenario illustrates the feasibility and benefits of our proposed framework.