Personalized rehabilitation can improve educational and clinical outcomes for children with special educational needs and disabilities (SEND), but creating individualized reading exercises, comprehension questions, lexical supports, and progress reports remains time-consuming for clinicians. We present ARTIS, a therapist-in-the-loop and natural-language-processing (NLP)-assisted platform that supports reading-comprehension rehabilitation while preserving expert oversight, and evaluate its deployment in a real clinical setting. We report a ten-session intervention involving ten children aged 9–11 years at a speech and language rehabilitation centre. Across the intervention, platform comprehension accuracy remained relatively stable, while pre/post analyses showed stronger gains in vocabulary knowledge and lexical competence than in standardized reading-comprehension scores. Changes in platform accuracy were positively associated with changes in standardized comprehension outcomes, although these exploratory results should be interpreted cautiously. Overall, the study demonstrates the feasibility of integrating therapist-supervised NLP into a real rehabilitation workflow and highlights its potential to facilitate personalized material preparation while maintaining clinician oversight.