This study examines how AI-enhanced Computer-Assisted Language Learning (CALL) tools can improve English language learning among autistic students in Pakistani inclusive classrooms. The aim is to examine whether adaptive AI characteristics—such as individualized pacing, multimodal interfaces, and computer-generated feedback—can facilitate engagement, comprehension, and emotional regulation among learners with autism spectrum disorder, while also benefiting neurotypical learners. The study adopts a small-scale mixed-method intervention design conducted in an urban Pakistani ESL context with 30 participants, including autistic and neurotypical learners. AI-based language learning applications were implemented over a period of eight weeks. Quantitative data were collected through pre- and post-language assessments, while qualitative data included classroom observations, teacher interviews, and parental feedback. The design incorporated ethical considerations and cultural relevance. The findings indicate improved vocabulary retention, reduced anxiety during language activities, and greater learner independence among autistic students. Teachers also reported that AI-supported tools did not increase instructional workload and supported differentiated instruction in inclusive classrooms. This study contributes to the field of CALL by highlighting AI-driven inclusive language learning within a Global South context. It offers practical implications for designing inclusive CALL environments, policy recommendations for integrating special education within mainstream ESL classrooms.