Dementia is associated with progressive cognitive decline, and its global prevalence is increasing, driving interest in assistive technologies to support everyday life. Recent work emphasises that co-design should begin early and continue through implementation, and that AI innovation in dementia care must be guided by person-centred principles. However, many technologies remain caregiver-focused, with limited integration of the perspectives of people living with dementia. This mixed-methods study explored views on AI-enhanced assistive technologies across five groups: people with dementia living at home and in residential aged care (interviews), and informal caregivers, care staff, and healthcare professionals (surveys). Participants rated reminder-based scenarios (medication, hydration, social, and leisure) presented in manual, adaptive (AI-enhanced), and monitoring formats. Outcomes included likelihood of use and concern about errors. Findings showed a high likelihood of using assistive technologies across all groups, with a preference for manual over AI-enhanced versions. Monitoring features were valued by caregivers and professionals but often perceived as intrusive by people with dementia. Concerns about errors were highest for medication and lowest for leisure. Differences between the perspectives of people living with dementia and other stakeholders, especially in hydration and social contexts, reinforced the importance of actively involving people with dementia in shaping technologies intended to support them. These findings highlight the need for person-centred, ethically grounded AI innovation to support ageing well with dementia.