Autism is a complex neurobiological disorder that affects brain development and manifests itself primarily in difficulties in communication and social interactions. Autistic people may also experience a restricted set of interests and repetitive behaviors. Among the promising tools supporting autistic people, systems supporting diagnosis, planning therapy, monitoring progress, and even conducting direct interactions with autistic children are becoming increasingly important. The article describes an expert system created to support the rehabilitation of children with autism. The presented expert system, for instance, can analyze the child's speech patterns and suggest appropriate communication strategies, or it can monitor the child's behavior and propose activities to address repetitive behaviors. The system is part of a more extensive solution consisting of three main components: the patient interface, the rehabilitator's expert system, and the machine learning module. The presented environment was used during therapeutic sessions with young patients. On the one hand, it provides promising results in patient interactions; on the other hand, it supports the work of rehabilitators. It is worth noting that the main strength of the solution is the inference and suggestions proposed by the system during a session with the patient; taking into account the dynamic development of artificial intelligence systems and the fact that each subsequent session provides new data, we can count on the beneficial impact of the presented system on both the progress in the patient's rehabilitation and also the quality and effectiveness of the rehabilitator's work.