Patent Pending · AU2026903561 · Melbourne 2026

Your emotions
decide before
you do.

EMPATHIQ is a patented emotion-aware agentic AI that continuously monitors physiological and behavioural signals, learns each person's unique emotional patterns, and intervenes — gently — before emotionally-driven decisions are executed.

Explore Collaboration → View Technology
75K+
Consumers in meta-analysis confirming emotion drives impulsive decisions
$20B
Projected emotion AI market by 2034 at 22% CAGR
20
Patent claims filed — DEH, RDRS, Pattern Mirror, on-device privacy
5
Integrated system layers — the combination not in any prior art
Novel Claim 01
Decision-Emotion History
The world's first system linking real-time emotional states to personal decision events and their outcomes over time.
DEH Engine
Novel Claim 02
Rash Decision Risk Score
Personalised LSTM trained on your own decision history. Not population norms — your patterns only.
RDRS · Scalar [0,1]
Novel Claim 03
Pattern Mirror
LLM shows you your own decision-emotion history at the moment you're about to decide. Not in any prior art.
LLM · Non-prescriptive
Architecture
100% On-Device Privacy
All biometric data stays on the user's device. Only differentially private gradient updates transmitted.
Federated + DP
The Problem

We make our worst decisions when we feel the most.

"The same person, returned to emotional baseline, will frequently reach a different — and better — decision."

High arousal emotional states — stress, excitement, loneliness, anger — systematically impair the prefrontal functions responsible for deliberative reasoning. No existing system links an individual's real-time emotional state to their personal decision history and intervenes proactively.

  • No system currently builds a personalised decision-emotion history for each user
  • Prior art emotion AI serves advertisers and employers — not the individual
  • Existing wellbeing apps are reactive — they respond after, not before
  • Population-level emotion models fail for individual-level prediction
See the Full Evidence Base →
231
Meta-analysis · JAMS · 75,000+ consumers
High-arousal emotional states confirmed as primary triggers of impulsive purchasing. Self-control and mood state are the key mediating mechanisms.
51
Meta-regression · PMC 2023 · 14,957 participants
Emotion-related impulsivity positively predicts risky decision-making across age, gender, and clinical vs. non-clinical populations.
90%+
Classification accuracy · Multimodal biosignals · 2025
Multimodal emotion classification in real-world settings now exceeds 90% accuracy — confirming technical feasibility of always-on monitoring.
The Technology

Five layers. One integration
no prior patent has achieved.

Full Architecture →
01
Sensing Layer
Multimodal Emotional State Sensing
HRV, EDA, BVP, skin temperature from wearables · Voice acoustics · Keyboard patterns · Context signals. Fused on-device into Emotional State Vector — Valence × Arousal × Dominance.
Layer →
02
Baseline & DEH · Novel Claim
Personal Baseline & Decision-Emotion History
Each user's unique emotional fingerprint built from their own data — not population norms. The DEH links every decision event to the emotional state that preceded it and the outcome that followed.
★ Novel
03
Risk Engine · Novel Claim
Rash Decision Risk Score (RDRS) Engine
Personalised LSTM model trained on each user's own DEH. Computes a real-time risk scalar [0,1] — probability this person will regret this decision — based on their own historical patterns only.
★ Novel
04
Intervention · Novel Claim
Agentic Intervention Engine & Pattern Mirror
Tiered interventions from a soft nudge to the LLM-generated Pattern Mirror — showing the user their own decision-emotion history at the moment of a detected decision event. Always dismissible.
★ Novel
05
Wellbeing Layer
Personalised Wellbeing Support Engine
ESV-matched content recommendations, weekly longitudinal emotional health reports, clinical escalation pathway. All data processed on-device — no raw biometrics ever transmitted.
Layer →
Collaboration

Seeking world-class partners
across four domains.

EMPATHIQ requires partners who bring research rigour, hardware capability, platform scale, or institutional support. A full technical briefing is available under NDA.

🔬
Academic Research
Research Institutions
The world's first dataset linking continuous physiological emotional data to real-world decision events and outcomes does not yet exist. We need empirical research partners using EMA methodology.
Enquire about research partnership →
🏢
B2B Partnership
Enterprise & Platforms
EMPATHIQ's decision-protection capabilities can be embedded into financial wellbeing, digital health, consumer banking, and workplace wellness programmes as an API or SDK.
Enquire about B2B licensing →
⌚
Hardware Manufacturing
Wearable & Biosignal Devices
Hardware partners achieve meaningful product differentiation as the sensing layer of a patented decision-protection system. We need HRV, EDA, BVP from wrist-worn devices.
Enquire about hardware licensing →
🎓
Industry PhD / ARC
University Partners
The EMPATHIQ research programme is structured for an ARC Linkage Project or Industry PhD, providing the empirical foundations for clinical and commercial validation.
Enquire about academic partnership →
Intellectual Property

Patent pending.
Priority date: April 2026.

Filed with IP Australia under the Patents Act 1990 (Cth). Application No. 2026903561 — establishing an international priority date recognised across 170 countries.

Research foundations openly published on this site. Full patent specification, blueprints, algorithms, and claims available to qualified partners under NDA.

Request Technical Briefing View Architecture →
🛡
Patent Pending
AU2026903561 · IP Australia · April 2026
CLAIM 01
Decision-Emotion History (DEH) — Links each decision event to the user's emotional state preceding it and its outcome. No prior art does this.
CLAIM 02
Rash Decision Risk Score (RDRS) — Personalised LSTM on the user's own DEH. Not population norms — this person's patterns only.
CLAIM 03
Pattern Mirror — LLM-generated natural language presentation of the user's own DEH at the moment of a decision event. Not in any prior art.
CLAIM 04
On-Device Federated Learning — All biometric data on-device. Only differentially private gradient updates transmitted.

20 total claims · 3 independent · 17 dependent · Full claims under NDA