EMORA
01 / 11 · Cover

EMORA

See the emotion coming,
in time to get it right.

Emotion-sensing hardware for families — starting with ADHD (6–12)

An embodied agent that senses the child in real time,
and warns you before emotions escalate — not after.

Homework at dusk: the child writes, the parent watches calmly — the EMORA desk robot, the parent's watch and the app all present in one quiet moment

EMORA · Sense — Intervene — Learn

The Problem

They lack a decision window.

A typical evening · homework tutoring
Homework at night: a mother beside her frowning child at the desk
The child starts resisting
↓
Subtle physiological & behavioral shifts — the decision window opens
↓
Arousal climbs rapidly
↓
The parent notices the conflict
↓
Too late — the moment has passed
Today's tools · each missing a piece
MedicationLong-term physiological managementNo real-time family layer
ParentingTeaches the right methodsFails in the heat of the moment
TherapyStructured interventionCannot always be present
SensorsDisplay physiological dataDon't tell parents what to do

“It's not that we don't know the methods — it's that in those few seconds, nothing comes to mind.”

— Beta family, interview record
The missing layer is not more information. It is real-time decision support inside the family.

The Wedge · Why ADHD First

Why ADHD first?

01

High-frequency

Homework, wake-up, meals, bedtime — conflict scenarios recur every day

02

High-pain

One misread interaction can escalate fast into parent–child conflict

03

Time-sensitive

The window where intervention works may last only tens of seconds

04

Measurable

Alert — action — outcome: every stage is observable

05

Longitudinal

The same family keeps producing interactions and data over time

The same environment that creates the problem also creates the training data.

The Product

From sensing to action

Child side: the real EMORA device — cat-eared head, softly glowing eyes — watching over a child's desk
Child Side

Contactless sensing

A millimeter-wave radar reads breathing, heart rate and movement; a microphone array catches crying and shouting. Nothing to wear, nothing to operate.

Parent side: the watch face and its real escalation alerts — from a gentle fluctuation notice to an urgent cry alert
Parent Side

Just-in-time alerts

A quiet watch vibration: “This may be escalating.” It appears only at the critical moment — fully silent otherwise.

App: two real screens — a calm, communicable state on the left, a high-risk state with a concrete what-to-do-now suggestion on the right
App · Decision Support

Actionable guidance

Not a data dashboard — it answers “what to do right now”: lower the ask, stop pressing on who's right.

Child
→
Sense
→
Predict
→
Parent
→
Act

Alerts reach the parent's wrist ≥20s before eruption · End-to-end latency P95 ≤2.5s

The Agent Loop

A closed-loop agent,
not a chatbot.

01

SENSE

Continuously ingests multimodal signals from the real world

→
02

REASON

Estimates escalation risk against this child's own baseline

→
03

ACT

Triggers the parent alert with actionable guidance

↑

Every real-world intervention is both an execution loop and a learning loop.

↓
06

IMPROVE

Makes the next judgment know this child a little better

←
05

REMEMBER

Stores triggers, timing patterns and intervention history

←
04

OBSERVE

Captures the parent's response and what follows

Embodied AI

AI that operates in the physical world

Digital AI

Waits for a prompt

Prompt — waiting for input
↓
Response — generating an answer
↓
Conversation ends

Context resets with every session — blind to the environment, accountable for nothing

EMORA · Embodied Agent

Lives in the loop

Sense — the real environment
↓
Act — on real outcomes
↓
Learn — from the result ↺

Always present — actions change outcomes, and outcomes sharpen the next judgment

Physical sensing

Input is a continuous stream from the real environment, not text typed by a user.

Persistent context

Learns this child's baseline — not a population average.

Real-world action

The output is not a paragraph; it is a parent's action at a specific moment.

Feedback

What happens after the action becomes input to the next judgment.

EMORA is building an AI agent that acts inside the loop of real family life.

The Data Moat

The moat is not accuracy.
It is the corpus.

What public data can tell you
“This child was anxious at that moment.”
What EMORA is accumulating
“Which signal combinations precede escalation; what the parent did; how it turned out.”
Real-world Intervention Corpus
Physiological / behavioral signals
↓
Estimated emotional state
↓
Intervention action
↓
Parent feedback
↓
Outcome

Text and dialogue are everywhere online. Corpora of signal → intervention → outcome can only grow inside real families.

The Flywheel
More families
More interactions
Outcome data
Personalization
Better interventions
Higher retention
↺The Flywheel
The moat is not fully built yet. The moat-building loop has started.

Traction

Real families,
real validation

Gen 1 · Wearable

Wearable

Discovery: kids refused to wear it — adherence is a hard constraint, not a parameter to tune

→
Gen 2 · Contactless

Contactless

Change: sensing moved into a desk device — zero burden, zero operation for the child

→
Gen 3 · Dual-end

Dual-end system

Change: decision support moved to the parent — the intervention finally closes its loop

wearable → contactless → dual-end → closed-loop intervention agent

Early-warning effectiveness
0%

Effectiveness ±5% — delivered ≥20s ahead, confirmed effective by the parent

Beta families
5

Beta families in continuous use

Intervention events
~200

Real intervention events, recorded

Longitudinal logs
~4wks

Weeks of longitudinal family logs

300+ ADHD families reached organically 20–30/wk inbound inquiries ¥0 paid acquisition 15 refundable pre-orders XbotPark hardware bootcamp · 2nd prize

The product definition changed because real families told us what was wrong.

Why Now

Three curves
are converging

01 · Edge AI

Continuous on-device
compute becomes possible

Small models and on-device inference make 24/7 local processing viable on consumer hardware.

02 · Low-cost Sensing

Contactless sensing
goes consumer-grade

mmWave-class sensing hardware keeps getting cheaper — home deployment becomes economically viable for the first time.

03 · Agentic AI

AI moves from recognition
to decision & action

From answering “what happened” to answering “what to do next”.

5 years ago Sensing was expensive
→
Then AI could recognize
→
Now AI can decide & act

Physical sensing × AI reasoning × Real-world action

Where the three curves converge — there is EMORA.

Why Us

Founder-market fit
× full-stack execution

Yuan Weiqing

Founder & CEO

  • 10+ years in product at Tencent / NetEase, P8+ product lead
  • Shipped 100M-DAU products · ran ¥100M+ revenue businesses
  • Hands-on 0→1 with AI agents and multimodal models
  • Parent of an ADHD child — I live this problem every day

Zeng Weifeng

Hardware Partner

  • Wuhan University · 15+ years in hardware engineering
  • Smart wearables and connected devices
  • Full chain: prototype → engineering → mass production
  • Execution on the Shenzhen supply chain
Product definition
→
Multimodal AI
→
Physical sensing
→
Hardware
→
Shenzhen production

Wearable → Contactless

Reality killed the hypothesis

“Kids will wear it” didn't hold — adherence is a hard constraint. We rebuilt the sensing architecture.

Single-end → Dual-end

Users redefined the product

The one who needs decision support is the parent, not the child. Action moved to the parent side.

Display → Actionable

The product became an agent

“Knowing” isn't enough — answer “what do I do now”, directly.

Our advantage is not getting everything right the first time. It is learning faster from reality.

The Vision

ADHD is our wedge.
Emotional intelligence
is the platform.

Today
One child · one family
one intervention window
Tomorrow
Thousands of families
one continuously learning agent

Not a better parenting app.
An AI agent that participates in real family life.

It starts with the few seconds when there is still time to change what happens next.

EMORA — Sense · Intervene · Learn

Night has settled: parent and child share a calm bedtime moment, EMORA glowing softly on the nightstand

EMORA · Thank You