With the recent technological advancements, new learning methods have emerged. Among them, two approaches stand out: immersive learning and AI-assisted learning. In industrial training courses designed to teach the operation of complex machines, it is very difficult to reproduce all possible situations while ensuring a high level of safety for learners. Immersive learning, which is based on extended reality (XR) technologies, makes it possible to simulate such scenarios in a controlled and safe environment. AI-assisted learning enables the automatic generation of training courses while providing personalized support to learners. Despite their potential, these two approaches present several limitations. Immersive learning requires significant investments in terms of time spent on course design, as well as technical expertise to build the 3D environment. Adoption remains limited due to user resistance to this type of technology. As for AI-assisted learning, it fails to replicate the essential physical and visual interactions required to operate industrial machines. This article presents an innovative framework that combines AI and XR to automate the creation of immersive training programs. Designed to address the limitations of current learning methods, the framework relies on an agent-based processing pipeline capable of transforming a single video recorded by an instructor into a complete training module. This module includes a multiple-choice test to assess the learner’s knowledge. A retrieval-augmented generation (RAG) system is also integrated, enabling real-time interactivity and providing learners with personalized answers to their questions.