The intolerance to frustration in individuals with autism leads to significant problematic behaviors with evident repercussions on their quality of life and learning processes. To overcome this problem, we create AIDA (Artificial Intelligence & Domotics for Autism). The recipients of the intervention for the implementation of AIDA are non-verbal autistic individuals who need to enhance their ability to make requests using images and have strong difficulty tolerating rejection. The AIDA is founded on the Hanley's “No” tolerance protocol: one of the most effective clinical methodologies to increase frustration tolerance levels. The computational implementation of AIDA is a distributed system which utilizes advanced technologies to manage user interactions, process requests, and provide real-time personalized responses. The Raspberry Pi has been selected as the central processing unit for its versatility, compact size, and affordability. The Random Forest algorithm is used for the classification of user requests and the generation of initial responses, providing a solid basis for managing interactions. The Random Forest predictions serve as the default response to user requests, enabling quick and accurate responses. Simultaneously, a rule-based algorithm is employed to handle specific situations and apply customized rules under certain conditions. This system allows for greater flexibility and adaptability, enabling responses based on specific contexts or user needs. The hybrid approach combines the predictive capabilities of machine learning with the predictability of rules, offering a sophisticated solution to enhance the quality of life for users with autism.