Warmth before efficiency
We care more about a screen feeling looked-after than about shaving taps. Tone is a feature.
About Alz SpaceDementia is not a solvable
Dementia is not a solvable
problem — but the daily experience is designable.
We're building the first platform that treats patient engagement, caregiver operations, and family participation as a single connected system — with the family's own memories as the raw material.
Why this exists
A household problem, not a patient problem.
More than 57 million people worldwide live with dementia. In the U.S. alone, seven million Americans are living with Alzheimer's today, with 200M+ family members directly affected and over 30 million active caregivers carrying the daily weight. Dementia does not live inside one person. It reshapes entire households.
Every tool we looked at had picked a single lane — brain-training apps that treat the patient in isolation, clinical rehab tools that don't fit the home, medication trackers and group chats that help with logistics but ignore the in-session experience. None of them serviced the household.
Alz Space is the attempt to treat engagement, operations, and family participation as one connected problem — and use the family's own memory material to solve it.
One engine behind three surfaces.
Caregivers assemble a daily timeline from a library of dementia-friendly modules — games, reminiscence photo playlists, music sets, companion conversations, calming rituals. An AI companion runs sessions, adapts pacing and difficulty in real time, and gently transitions between modules.
Underneath, a cloud LLM (Claude / GPT) sits behind a model-abstraction layer so we can route each AI job — schedule building, in-activity companionship, adaptive difficulty, report generation — to whichever model best fits cost, latency, and privacy constraints for that job. The three surfaces draw from a single engagement record and memory library; the same architecture maps cleanly onto care institutions, with a staff caregiver in the primary caregiver role.
Inputs
- Engagement history
- Family memory content
- Caregiver notes & ratings
Personalization engine
The top-priority AI workstream — learns every household, feeds every other job.
Outputs
- Schedule proposals
- In-session companion
- Adaptive difficulty
- Weekly report
Three principles we hold to.
Rhythms, not tasks
We design for the day a family actually lives — morning rituals, afternoon dips, evening wind-downs — not a to-do list.
Dignity is the product
The patient's surface is an engagement surface, not a surveillance one. Family content reaches the patient only through the caregiver.
Three phases, U.S.-first.
The product roadmap is depth-before-breadth. The AI layer is architected to keep changing underneath without changing the product on top.
- Phase 1
MVP to production
Finish the v1 platform — module library, three interfaces, companion, scheduling modes, memory library, weekly report — and harden for reliability and scale.
- Phase 2
U.S. East Coast pilots
Boston and New York first, mixing at-home households with institutional partners. Every pilot site is a validation unit feeding product iteration.
- Phase 3
National expansion
Depth before breadth. We market nationally only once the pilot markets have demonstrably changed the texture of a family's day.
AI evolution
- The personalization engine is the single biggest near-term investment.
- Real-time adaptive session intelligence — pacing, difficulty, graceful transitions.
- Weekly-report generation treated as a first-class AI product.
- Per-region, per-job model swapping as regulations and capabilities shift.
Privacy
We care about your family's privacy.
Dementia care involves some of the most sensitive material a family ever shares — photos of a parent, voice recordings of a spouse, daily notes. We store it carefully, the primary caregiver gatekeeps what reaches the patient, and we never sell data or run advertising. The full details live in our privacy policy.
Read the privacy policy