◉
Multimodal models got good enough
Vision-language models are finally reliable enough to turn a raw frame into an evidence-backed memory — not just a caption.
The problem
Everyday memory gets harder. Family isn’t always in the room.
As people age, small everyday memory lapses — where did I put this, did I already do that, what happened earlier — get more frequent, and more disruptive. The people who’d normally help, adult children and other family, aren’t always there when the question comes up.
Today’s AI glasses can see and answer in the moment, but that isn’t the same as a memory. Once the moment passes, there’s no trustworthy record to check later, and no easy way to bring a trusted person in when it calls for more than a quick answer.
MEMO
Visual memory, a remote call, and day-to-day AI — one local system.
Tap a card to see how it actually works.
Visual Memory
Evidence, not footage
Confirmed sightings of real objects and places become a durable, queryable memory — built from bounded evidence, not a video archive.
Outcome: a memory you can ask, and trust the answer.
Tap to see how it works ↻
How it works
Seen
→
Confirmed
→
Remembered
A confirmed sighting becomes a durable memory, with its evidence attached — not a video file.
Remote Call
Human in the loop
When judgment matters more than an automated answer, MEMO opens a live call to a trusted person, seeing what the wearer sees.
Outcome: human judgment stays within reach, exactly when it matters.
Tap to see how it works ↻
How it works
Wearer
→
Live call
Family
One request opens a live call so a trusted person can see and hear what the wearer does.
Day-to-Day AI
Hands-free, glasses-native
Everyday questions, answered locally, without pulling out a phone or repeating context the system should already have.
Outcome: assistance that fits into the day instead of interrupting it.
Tap to see how it works ↻
How it works
Ask
→
On-device
→
Answers
No cloud round-trip — the question is heard, reasoned over, and answered on the same local hardware.
How it works
Event-triggered evidence, understood once, remembered, retrievable later.
1
Event-triggered evidence
→
2
Understanding
→
3
Memory
→
4
Retrieval
MEMO doesn’t record continuously. A meaningful event triggers a short capture, a model turns that capture into an evidence-backed memory, and later a question retrieves it — with the evidence to back it up.
Privacy thesis
Evidence, not surveillance.
A useful memory does not require continuous recording. MEMO’s premise is that a short, bounded clip of a confirmed event is enough to make an answer trustworthy — and that anything beyond that evidence window is a liability, not a feature. No rolling video of your day, no footage of the people around you, no record kept because it might someday be useful.
Why now
Three curves crossed at once.
◎
Smart glasses became wearable
Camera-equipped glasses like the RayNeo X3 Pro are light and unobtrusive enough for daily use, not just a demo.
⌖
Edge AI hardware caught up
Compact local hardware, such as the Acer GN100, can now run perception, speech, and reasoning together, without sending anything to the cloud.
The thesis
AI glasses will need memory, not just intelligence.
Today’s assistants mostly reason over the present moment. A wearable assistant becomes substantially more useful when it can remember the physical world over time. MEMO is building that memory layer:
1
Perception
→
2
Evidence
→
3
Memory
→
4
Retrieval
→
5
Action
Not just an interesting smart-glasses app for older adults — a visual memory layer for wearable AI, with older adults as the first market.
What exists today
A working prototype, not a pitch deck.
MEMO runs end to end today, on real hardware, in public.
Working prototype
RayNeo X3 Pro glasses paired to a single Acer GN100, running perception, speech, reasoning, and memory entirely locally.
Hackathon winner
Winner, NVIDIA Spark Hackathon Seattle 2026, judged on a live, working system.
Public demo
A recorded demo and a live in-person walkthrough are both available on request.
Runs locally
The current prototype runs locally, keeping visual and personal context on local hardware.
Where we start
Adults and older adults who want to remember, stay connected, and get everyday help.
MEMO starts with adults and older adults who want help remembering everyday things, staying connected to family, and getting contextual assistance through smart glasses.
⌖
Memory
“Where did I leave my keys?” “Did I already do this?” “What happened earlier?”
◎
Connection
Call a family member or caregiver directly from the glasses when help is needed.
◉
Daily assistance
Ask what to cook, what an object is, what to do next, or get contextual guidance.
MEMO is not designed to continuously record a person’s life. It creates selective, evidence-based memories when useful events occur.
Who might pay
Who might pay?
One hypothesis is that the wearer is the user while an adult child or family member may be the buyer. If that holds, it would matter for the business model — but like the questions below, it’s untested.
Start
Direct-to-consumer
→
Then
Family-supported adoption
→
Later
Senior living / care organizations
What we need to validate
Four open questions, before any market claim.
Customer problem
Is the memory and connection gap we’re solving actually a top-of-mind problem for older adults and their families, or a nice-to-have?
Workflow fit
Does glasses-based use fit naturally into daily life, or does it add friction that outweighs the benefit?
Willingness to pay
Will a family pay for this, and at what price relative to what they already spend on care and safety tools?
Deployment model
Hardware, setup, and support — what does it take to actually get this into someone’s home and daily routine?
These are open questions, not results. We have not run paid pilots, signed customers, or generated revenue — and won’t claim otherwise.
Roadmap
From working prototype to validated pilot.
Now
MVP hardening
→
Next
Customer discovery
→
Then
First pilot
→
After
Measured results
The team
The people building MEMO.
Reach any of us directly to learn more.
Message any of us directly about what we are building.
Investors & partners
Building MEMO’s next chapter.
If evidence-based visual memory for older adults, their families, and the care organizations that support them is interesting to you, we’d like to talk.
Alexander Kuznetsov
Investor & partner contact