Follow-up systems for vocational training, particularly in work-study schemes, are essential to ensure that apprentices make progress in acquiring specific skills. This article presents a conversational tool integrated into a digital training booklet. The tool combines a knowledge graph, linked to the skills to be validated, with large-scale language models (LLMs) and a generative model based on BERT. This approach enables intelligent management of learning data while allowing natural language interactions to monitor progress and provide personalized coaching. By offering personalized feedback and advice, this tool aims to improve the support, ensuring more effective skill acquisition. It offers a more adaptive and interactive way of tracking progress, making it a key innovation in the field of vocational training.