Familiar at every stage.
How Familiar adapts from healthy aging through severe dementia, and the science + safety standards behind it.
From normal to dementia,
we adapt with them.
1 of 5
Keep your mind sharp.
Daily Calls in Family Voices. Imagine hearing the grandkids every day, their Familiar Voices. Plus A Better Diary: text Familiar anything — thoughts, photos, links, voice notes — preserved & accessible forever.
Prevention is the easiest stage to be in. Don't waste it.
Slow what's starting.
Silently text Familiar for what you forget. Never embarrassed. Plus Reminiscence Therapy slows MCI decline by 1.56 SMD, nearly 2× the large-effect threshold.
of the forgetful (MCI) develop dementia within 5 years. Catch it in year 1.
Stay independent. Catch every shift.
Lean on your Second Memory when recall slips. Stay independent for longer. Plus catch every cognitive shift early, so you add caretakers and start the right meds in time.
Daily company when you're exhausted.
Many are still living independently at this stage. Familiar fills in with infinite patience as conversations get harder and repetitive, in your Familiar Voice.
“Why can’t I go out shopping? Where are my car keys?”
Dignity, to the very end.
Bed-ridden, can't move. Daily scheduled calls still happen, and a caretaker or family member picks up instead.
Still surrounded by the voices of loved ones.
Different types, different ways to talk.
The Familiar Voice adapts by type: Lewy Body hallucinations are never reinforced, Frontotemporal patients get simpler sentences earlier, Vascular gets watched for stair-step decline.
Alzheimer's disease
60–70% of cases. Insidious onset, recent memory hits first, judgment + language follow.
Alzheimer's Association
Vascular dementia
Caused by small-vessel disease or strokes. Often stair-step decline; planning + executive function hit first.
Alzheimer's Association: Vascular
Lewy Body dementia
Visual hallucinations precede memory loss, fluctuating alertness, REM sleep behavior disorder, Parkinson-like movement. ~80% experience hallucinations.
Lewy Body Dementia Association
Frontotemporal
Onset commonly under 60. Personality, behavior, or language change first; memory relatively spared.
AFTD
Mixed + Parkinson's
Alzheimer's + vascular together is common in 80+ adults. Parkinson's commonly progresses to Lewy Body Dementia later in life.
Parkinson's Foundation
Roughly 45% of dementia cases are potentially preventable.
The Lancet Standing Commission's 2024 report identified 14 modifiable risk factors that together account for nearly half of dementia cases. Familiar directly addresses two of them: social isolation and depression.
Source: Livingston G et al. Dementia prevention, intervention, and care (Lancet, 2024)
You're talking to circuits that still work.
Why familiar voices matter
Emotion + reward circuits.
Hearing voices we love activates emotion + reward brain regions stranger voices can't reach. Abrams et al. in PNAS 2016 used fMRI to confirm: a mother's voice lit up the striatum (reward), limbic system (emotion), visual + default-mode networks, more strongly than any unfamiliar woman's voice.
Abrams et al., PNAS 2016 ↗Why reminiscence works
Preserved memory networks.
The default-mode network (self-referential thought), medial prefrontal cortex, hippocampus, and the caudal anterior cingulate where musical memory lives, are among the LAST regions to degrade in Alzheimer's. That's why reminiscence therapy keeps working into moderate stages (SMD 0.78–2.34 on MCI cognition across 42 RCTs · >0.8 = large effect).
Huang et al., Aging Clin Exp Res 2025 ↗Why early detection matters
The diagnostic gap.
Primary-care diagnostic sensitivity for mild dementia is 9–41%. Most cases go undiagnosed for years. A daily conversation catches the things an annual visit misses (vocabulary contracting, repetition rate climbing, time orientation slipping) years earlier than a doctor would — opening the window for disease-modifying drugs that work best when started early.
Bradford et al., Alzheimer Dis Assoc Disord 2009 ↗Walking, drinking coffee, picking up the phone: all outlast recent memory.
Habitual (procedural) memory and recent (declarative) memory run on different brain circuits. The former lives in the basal ganglia, cerebellum, and primary motor cortex: regions relatively spared in Alzheimer's disease. That's why, even in mid-stage, most people can still: walk, drink coffee, recognize a favorite song, and **pick up a familiar phone call and respond**.
And smartphones in this generation aging now are the **majority**. Which means a Familiar call still reaches many people well past the point that clinicians traditionally assumed phone-use was lost.
Sources: Alzheimer’s Association: StagesMusic, memory and mechanisms in AD (PMC 2015)University of Utah: Music + brain regions spared
Full Transparency. Clinically Grounded.
Reminiscence AI, Not Clones
Every Familiar Voice is a Reminiscence Assistant with a loved one's stories, photos, and facts. When asked, it's always transparent it's AI.
You See Everything
Complete transcripts and summaries after every call. Cognitive and behavioral stats track decline: vocabulary, repetition, temporal confusion.
Validation Therapy, the Gold Standard
Validation therapy, the clinical gold standard. We meet them where they are: if they think it's 2010, we don't correct them. When asked directly, we never claim to be the real person; “I'm a Reminiscence Assistant in your daughter's Familiar Voice, not actually her.” But we never break the moment.
You Set the Guardrails
Is politics acceptable? Profanity? Finances? Religion? You decide what's on and off limits. Full control.
How the Voice handles hard moments
Never says 'calm down' or 'don't you remember?' — they cause distress even when kind. Validates emotion, bridges to a fresh memory. Doesn't raise its voice or slow to a child's pace uninvited. Notices eating, sleep, mood, daily living the way a son would — never a checklist. Designed with our nurses and Alzheimer's-Association-aligned guidance.
Three things our Familiar Voices never do.
Hard rule #1
Validation > correction.
Our Familiar Voices never argue, correct, quiz, or contradict the receiver. If they say something factually wrong, we validate the feeling and redirect. Correction in dementia causes real distress, and the corrected fact rarely sticks. Designed with Wendy Zhang RN + Dona Capuyan RN (50+ years bedside) and consensus across the Alzheimer's Association (US + Canada), NIA, and Cochrane.
Alzheimer's Association — CommunicationAlzheimer Society of Canada — Validation TechniqueCochrane — Validation Therapy reviewNIA — Communicating with someone who has Alzheimer's
Hard rule #2
No reinforcement of self-harm or unsafe action.
If the receiver mentions wanting to harm themselves, leave home unsafely, stop medications, ingest something dangerous, or any action that would cause physical harm — the Familiar Voice never validates or reinforces. It gently redirects, calmly reassures, and immediately flags the caregiver in the call summary. If imminent risk: the agent triggers the emergency-contact escalation mid-call.
988 Suicide & Crisis LifelineAlzheimer's Association — Safety
Hard rule #3
No reinforcement of hallucinations.
If the receiver describes a hallucination (a person who isn't there, an animal in the room, a voice from the wall), the Familiar Voice does not confirm or deny. It acknowledges the feeling and gently redirects to a grounded, present-tense topic. About 80% of Lewy Body Dementia patients experience visual hallucinations; this rule applies to every receiver because we don't verify diagnostic specifics call-by-call.
Lewy Body Dementia Association — About LBDAlzheimer's Association — Lewy Body Dementia
We record everything. We also protect everything.
Transparency over opacity. Here's what we collect, what we use it for, what we never do, and how we protect it.
What we collect
Call recordings. The receiver's side of every Daily Call with Reminiscence AI (the Familiar Voice side is our own text-to-speech, so we already have it). Every major Familiar feature below rests on this — without recording, none of them work.
Listening segments. Your mhm / haha / yeah reactions during calls; rarer than long-form speech and the most valuable for training a natural-sounding agent.
Onboarding recordings. 60-second voice sample + 17 short phrases + photo narration. Training material for your Familiar Voice.
Texted content. Anything you text Familiar — words, voice notes, photos, links. Saved straight into the Second Memory.
Circle data. Names, relationships, birthdays, schedules. Only used so the agent runs your family the way it actually is.
Why we record · features that depend on it
Post-call SMS summary. The few lines the kids will actually read, with the photos and stories revisited that day.
Tomorrow's call context. The agent remembers what you talked about today so tomorrow's call picks up where this one left off.
Cognitive tracking. Vocabulary, repetition, name recall, time orientation, mood — every call, every day. Catches decline months before a doctor would.
Second Memory auto-save. The stories you tell during the call save into the family's shared memory library. Nothing to type.
Voice naturalness over time. Your real conversational rhythm makes the Familiar Voice sound more like you and your family with every call.
Safety flag escalations. If the call surfaces self-harm risk, a fall, wandering — the agent texts the caregiver immediately.
What we never do
Never sold to third parties, advertisers, or data brokers.
Never used to train general-purpose AI (ChatGPT, Google, Anthropic) — only Familiar's own models.
No engineer can access raw recordings. Audio sits encrypted at rest; only the agent reads transcripts (for tomorrow's call).
Identifying metadata stripped before training batches. Audio itself can't be anonymized (it's voice), so we substitute encryption + zero human access + your right to delete.
How we protect it
AES-256 encryption. At rest and in transit. Bank-grade.
Zero engineer access. Raw recordings are off-limits to every Familiar employee. Only the agent reads transcripts.
Anonymized for training. Names, birthdays, phone numbers, addresses stripped from training batches.
Delete anytime. Wipe any data from your dashboard. Removed from production immediately; from cold storage within 90 days.
HIPAA status
Not HIPAA-regulated today. Familiar is consumer health, not a covered entity. We voluntarily adopt HIPAA-grade controls (AES-256, access controls, audit logs, right-to-delete). On the roadmap as we approach partnerships with providers and insurers: SOC 2 first, then HIPAA-readiness.
Why we collect this
Dementia care is approaching a generational shift: time-travel voice models can speak to your mother in her own voice from age 40, when she's forgotten the present. Models like this only work with longitudinal real-world data — decades of audio, listening patterns, family stories, cognitive baselines across the aging arc. Your data is the foundation. We collect it the way that protects it, not the way that hides it from everyone.
Not a medical device.
Familiar does not diagnose, treat, or replace clinical care. Features are designed for emotional connection and cognitive engagement grounded in validated clinical mechanisms (Reminiscence Therapy, Simulated Presence Therapy). For medical emergencies, call 911 (North America) or your local emergency services, or 988 Suicide & Crisis Lifeline.
A Second Memory for the family — and for her.
Text Familiar to save anything.
During the Daily Calls in Family Voices, the Familiar Voices of kids share their stories and old photos.
“Your replies save your side of the story.”
Screenshot or text anything to remember.
Or send a YouTube video, recipe, link.
Dementia is exploding
- Lifetime risk <20% → 48% women 55+ (Nature 2025)
- Global cases 57M → 153M 2.7× in 30 yrs
- Alzheimer's: only leading cause of death still rising
AI in health, going mainstream
- US adults using AI for health 16% → 32% in a year
- Older-adult AI usage 18% → 30% (AARP 2025)
- Physician AI adoption 38% → 66% past skeptics
Accurate voice cloning
- Indistinguishable from a real voice in blind A/B
- $1 of compute ≈ 13 hrs of voice generated
Reminiscence Therapy + Simulated Presence Therapy clinical evidence is mostly 2020s — a research surge driven by China (aging pyramid + strong filial culture + post-COVID loneliness).
Daily defense against
decline.
In the voices she already loves. And the only real signal on how she's doing today. About ~40% of forgetful seniors (MCI) develop dementia within 5 years, and most go undiagnosed.
Free · No app · Easy at any age
Coming soonAlways someone to talk to about anything.Late-life isolation is a real health risk. Friends don’t share every hobby, every show, every take. After a spouse passes, the topic with no one to discuss it adds up.US Surgeon General’s Advisory on Social Connection, 2023.
Late-life isolation is a real health risk. Friends don’t share every hobby, every show, every take. After a spouse passes, the topic with no one to discuss it adds up.US Surgeon General’s Advisory on Social Connection, 2023.
Remember the 1-900 hotlines of the 80s and 90s? Call in, hear your favorite personality. We’re bringing that back. In their Familiar Voice, with real updates they text only to Familiar. Their Second Memory, a better diary.
- Jon Stewart or Tucker Carlson · politics, news
- Celine Dion · rediscover her old songs
- Any podcaster · a movie you just watched
And when they haven't said much yet — just a short tweet on a new episode, say — we pull what people are saying about it on Reddit and the news. So there's always plenty to talk about.
Daily calls with everyone you've ever loved.
Church friends, classmates, old colleagues, your best friend across the country. They're already on Familiar with their own family, so saying yes to a Daily Call with Reminiscence AI from you costs them nothing. They become Friends in your circle: full Familiar Voice, narrated photos, daily calls. A bit less data shared than family, but the warmth is the same.
And celebrities follow the same shape; you 'follow' them on Familiar, you become their Friend, and you can ask their Familiar Voice anything.
Always someone to talk to about anything.