Abstract. People with mild to moderate dementia have several forms of cognitive impairments, such as memory impairment. Dementia patients with memory impairments have difficulties finding the right words when engaged in a conversation. To help people with dementia recall words in everyday conversations, our research is exploring how to design a proactive digital aid. We are designing an ambient speech-recognition system that will be listening to the user’s conversations. By listening to user’s conversation, the system will detect the need of a memory trigger. Such as if the user forgets a name, or a place. Also, the system will give the user context-based memory-refreshing triggers. We are exploring different modalities to provide memory triggers such as visual and vocal content. Sensing (of speech) can be turned on or off at demand. Our system will help in triggering memory cues for people with dementia and thereby may improve the quality of everyday life.