As digital services become a natural part of daily life, ensuring accessibility for people with cognitive disabilities remains a major challenge, despite laws and guidelines. This thesis explores how artificial intelligence (AI) can be used to support more accessible information architecture (IA) and improve digital inclusion for users with cognitive needs. Using a mixed-method approach, we combine a survey targeting individuals with cognitive variations, a semi-structured interview with an expert from the National Agency for Special Needs Education and schools (SPSM, in swedish), and practical testing of three digital tools: HeadingsMap, Eye-Able & HelperBird. These tools were evaluated based on their ability to detect and reduce common cognitive barriers, such as confusing layouts, unclear navigation and overwhelming text. The findings show that AI-driven tools can offer real support, simplifying language, clarifying structure and adapting content to individual needs. At the same time, they raise important concerns around data privacy, transparency, and the unintended risk of excluding users through over-customization. This thesis contributes to the growing field of accessible and inclusive design by providing practical insight nto how AI, when thoughtfully integrated, can complement human-centered practices. It's not about replacing people with machines, but about using technology to meet people where they are, and to help make digital services work better for everyone.