I’ve been thinking a lot lately about how we talk to our devices and what happens when those conversations become genuinely helpful rather than just transactional. Last month at a health tech meetup in the Mission District, I watched a demo of voice-activated medication reminders and thought we were still years away from AI that could truly understand context and adapt to cognitive decline. Turns out I was wrong.
A company called CrossSense Ltd just won a million-pound prize for developing AI software that embeds into smart glasses to help people with dementia navigate daily life. The Longitude Prize on Dementia, supported by organizations including Alzheimer’s Society and Innovate UK, recognized technology that could genuinely extend independence for the roughly 150 million people expected to live with dementia by 2050, according to the World Health Organization. That’s not a distant problem anymore.
What sets CrossSense apart isn’t just another reminder app disguised as wearable tech. The software creates an AI assistant named Wispy that lives inside chunky black-rimmed glasses equipped with a camera, microphone, and speakers. Wispy doesn’t just tell you to take your pills. It guides you through making tea, asks how you’re feeling, engages in actual conversation, and helps trigger memories. Information appears as floating text in your field of vision while a voice offers verbal cues, creating what researchers call multimodal support.
Szczepan Orlins, chief executive of CrossSense Ltd, told The Guardian that a smartphone version launches later this year. The full smart glasses experience should hit the market in early 2027. His team partnered with various hardware manufacturers to create frames compatible with prescription lenses and hearing aids, acknowledging that assistive technology needs to fit into people’s existing lives rather than demanding they adapt to clunky gadgets.
The science behind this matters. Professor Julia Simner from the University of Sussex led testing with twenty-three pairs of dementia patients and their caregivers. Without the glasses, participants correctly identified only forty-six percent of household items. With CrossSense running, that figure jumped to eighty-two percent. Here’s what caught my attention: an hour after removing the glasses, accuracy remained at seventy-eight percent. The technology appeared to reinforce cognitive pathways rather than simply compensating for their deterioration.
Dr. Foyzul Rahman, a cognitive decline expert at Loughborough University who wasn’t involved with the project, emphasized to The Guardian that CrossSense offers real-time prompts and feedback during tasks, not just calendar notifications. That distinction transforms the technology from a digital Post-it note into something closer to a patient companion who understands context and progression.
But Rahman also raised concerns I’ve heard echoed across the assistive tech community. Larger randomized trials need to verify these preliminary findings. Ethical questions around consent become thorny when devices collect continuous data from users whose cognitive capacity fluctuates. And there’s the practical issue plaguing many smart glasses: battery life currently maxes out at one hour, requiring users to carry portable power banks.
The business model targets consumers first with an expected fifty-pound monthly subscription plus up to one thousand pounds for the glasses themselves. CrossSense hopes the NHS will eventually cover costs, though that pathway remains uncertain given healthcare budget constraints across the UK and similar systems globally. According to MIT Technology Review, healthcare AI applications face significant adoption barriers around reimbursement structures and clinical validation requirements.
I spoke with developers at a recent AI ethics conference who pointed out that machine learning systems need enormous amounts of personal data to truly adapt to individual users. CrossSense employs this approach, with Wispy learning from each interaction to adjust support levels as conditions change. Caregivers can input information through a companion app about care needs and preferences. Over time, the system theoretically becomes more attuned to subtle shifts in cognitive function.
Carole Greig, a seventy-year-old woman diagnosed with Alzheimer’s nearly three years ago, tested the technology and described it to The Guardian as “amazing.” She highlighted something technical specs often miss: the emotional weight of maintaining independence and not feeling like a burden. That psychological dimension matters as much as whether someone remembers where they put their keys.
The broader context here involves a race among tech companies to crack the assistive AI market. According to Wired, several firms are developing similar augmented reality tools for cognitive support, though most focus on narrow applications like navigation or facial recognition. CrossSense’s integrated approach combining task guidance, conversation, and memory support represents a more ambitious vision of ambient intelligence.
Still, questions linger about real-world adoption. Studies published in the Journal of Medical Internet Research show that elderly users often abandon assistive technologies that feel stigmatizing or overly complicated. Smart glasses already carry social baggage after previous commercial failures. Will people with dementia and their families embrace wearing cameras and microphones throughout the day, even if the benefits are substantial?
There’s also the matter of technological equity. A thousand-pound device plus ongoing subscription costs puts this firmly in the luxury category for now. If these tools prove genuinely effective, access disparities could deepen existing healthcare inequalities unless public health systems step in.
What excites me about CrossSense isn’t just the technical achievement but the design philosophy. Too often, technology for cognitive decline treats users as problems to be managed rather than people to be supported. An AI that can joke around, help you remember your grandchild’s name, and guide you through making breakfast while reinforcing neural pathways feels fundamentally different from a blaring alarm reminding you to eat.
The technology landscape is shifting toward these kinds of contextual, adaptive systems. Whether CrossSense specifically succeeds matters less than whether it demonstrates a viable path forward. Because those 150 million people living with dementia by mid-century deserve better than institutional care as their only option when independence becomes difficult. If AI can genuinely help with that, we should pay attention.
