Specific Learning Disorder (SLD) is a neurodevelopmental condition affecting children’s abilities in reading, writing, or mathematics, and is often misattributed to low intelligence or poor education. Despite a growing body of evidence on the utility of technological interventions such as social robots in educational and therapeutic contexts, existing design frameworks for AI systems lack specificity when applied to human-robot interaction in SLD settings. In response, this study introduces and validates a novel Dual Pyramid framework tailored for human-centered AI-based robotic systems aimed at empowering the SLD community. A qualitative, multi-phase study was conducted to validate this framework, involving semi-structured interviews with 11 stakeholders, including children with SLD, their families, special education teachers, and clinical psychologists. Analysis of interview data revealed five core design priorities commonly emphasized by all stakeholder groups: effectiveness, safety, privacy, understandability, and explainability. These findings highlight the critical design parameters that should be prioritized when developing therapeutic robots for SLD contexts. Accessibility, although not among the top five prioritized needs, received a high importance rating, underscoring its relevance. Overall, this validation offers a practical guideline for future development and customization of robotic interventions for SLD populations and contributes to the broader discourse on ethical, inclusive, and effective AI in therapeutic contexts.