This paper presents the development of a digital generative design tool for residential building that integrates qualitative data from potential users of buildings. The central aim is to understand and challenge the inherent biases in the design process of architecture for mobility impaired users, whose experiences might be difficult to understand for designers who often move around and use buildings without any difficulty.

Although Universal Design promotes designed environments that are more sensitized with the diverse difference of individuals, the most of design generating tools are based on empirically deducted human needs, objectifying the people or seeing them as useful in simply validating design ideas. There is a clear distance in between the real needs and wishes of wheelchair users and what architects imagine when designing.

Mixed-methods – expert interview, literature review and data analysis of disability blogs – are used to collect and analyse wheelchair users’ experience. Accumulated qualitative data is, then, used as guiding input for the development of an explorative generative model that effectively produces large number of floor plans for residential architecture.

The developed generative model effectively selects floor plans that correspond with challenges described by the wheelchair users. The selected floor plans become informed starting points for spatial planning, which can guide architects to produce new and unexpected design solutions that are more sensitised to wheelchair users’ experience.

The application of generative design tools in early stages of design tasks can help architects to understand users’ needs and wishes, and thus challenge biased assumptions about wheelchair users’ experiences. And yet, further research needs to be conducted in order to progress the system: additional user data and new design objectives can give rise to new hypothesis and allow the system to be more precise, responding to the complex reality of disabled people in their everyday lives.