ALOS
(2026)
A digital peer-to-peer platform designed for caregivers of elderly with Alzheimer’s and other forms of dementia. Built through continuous dialogue with real caregivers, the platform directly addresses the isolation that follows a diagnosis and provides support for daily struggles. At the core of the system is an AI-powered matchmaking engine that connects caregivers seeking advice with peers who have navigated the same challenge. The platform also features a personal caregiver diary and a curated guide to local support networks, which frequently go unnoticed.
4 million people in Italy alone care for elderly with Alzheimer’s and other forms of dementia. However, following a diagnosis, family caregivers often navigate medical care, bureaucracy, and daily life entirely alone.
To understand caregivers' core needs and their attitudes toward AI, we conducted a four-phase user research. The findings reveal that caregivers feel abandoned by the healthcare system, forcing them to constantly improvise practical solutions on their own. Consequently, there is an urgent need for a reliable support network—a safe space to exchange advice with peers who truly understand their daily realities. While caregivers are open to AI helping build and facilitate this network, data privacy has to be guaranteed.
So, how might we connect caregivers with the right peer knowledge when they need it, to reduce isolation? Our solution is Alos. It's a digital service that helps caregivers access practical guidance by using AI to build meaningful, context-aware peer connections with those on a similar path. The platform also offers a personal diary, community data insights, and access to local associations.
Alos is the chatbot at the core of the service. When a new problem arises, the caregiver describes the problem in their own words, and the AI finds three people who had faced and solved the same problem. The output is three cards with the closest matches. After that, the relationship becomes entirely human: the caregiver can chat with the ones identified by Alos.
The diary is a space where caregivers can log moods, daily care routines, and private reflections. It answers the need of an emotional safe space. It maps emotional patterns over time, and uses past check-in data to fine-tune future recommendations.
The chat with the caregiver is a private 1-to-1 messaging channel for matched peers. The enclosed setting builds trust and is safe spaces for practical advice and emotional exchange. After some time after the chat, the app asks if the match was useful, so the solution is stored into the useful solutions database for the AI.
Once the caregiver helps, they gain specific badges to recognise and reward them. It values asking for help as much as offering it, while highlighting practical care expertise across the network.
As we discovered, caregivers struggle to find associations and pratical support in their local area. This section is a local directory mapping nearby caregiver associations, specialized care centers, and community initiatives.
To build the sense of community, a dashboard section displays collective data, shared activity, and community milestones. Visualizing peer metrics validates caregiving struggles and shows users they belong to an active network.
After downloading the app, the onboarding flow addresses three specific inputs: caregiver profile to capture personal workload and emotional baseline; care recipient context to detail dementia stage and daily behavioral patterns; privacy framing to clarify the user's primary motive for downloading the app.
The prototyping phase consisted of three phases. Each phase was informed by the testing's results of the previous one. The tests were conducted with caregivers to gain the most important insights, supported by peer testings and expert reviews sessions.
The AI model to find the best caregiver matches is RAG-Powered Recommender System. Semantic search quickly narrows candidates by lived experience, while an LLM delivers the final judgment on match quality.
Key learnings
Designing transparent AI-to-human handoffs;
Mixed-method UX research & synthesis;
Advanced component architecture & Figma prototyping;
Moderated usability testing & user facilitation;
Design system consistency & accessibility principles;
Additional credits
Diletta B. Canè, Cecilia Ferrentino, Agnese Pozzi, Vittoria Volontè