Emotion AI
labelled you.
EI understands you.
One was banned. Other is board-auditable.
For 10 years, everyone built Emotion AI that said "you are 78% angry". It got banned under EU AI Act Art 5.1(f) in workplace and education. We built EI that says "Conviction 42% LOW, Receptivity 78% HIGH — simplify, don't pitch". Same 5 signals — transcript + verbal + visual — opposite philosophy, opposite outcome, opposite regulatory fate.
78% Dashboard
after call
Emotional Intelligence
Attention MID
Adapt live
DIFFERENT PHILOSOPHY
Emotion AI tried to read your mind.
Science said you can't.
Ekman 1970s — 6 basic emotions. Happy, sad, angry, fear, surprise, disgust. Assumed face = emotion. Built Affectiva, etc. Said "your face shows 78% anger". Enterprise bought dashboards.
But 4 failures killed it:
You cannot label emotion from face alone.
But you can predict intent from pattern.
What research says — we are not making this up:
- Barrett 2019, Psychological Science: "People do not display one-to-one facial expressions for emotions. Same face, many emotions. Emotion categories are constructed by brain, not detected by face."
- EU AI Act Recital 44 + Art 5.1(f): "Emotion recognition in workplace and education is prohibited because it is scientifically contested and risks discrimination. Even if accurate, it is banned."
- IEEE 2022 Meta-Review: Emotion AI inter-rater reliability 0.28-0.44 — lower than chance for complex workplace scenarios. Not deployable for hiring, tutoring, meetings.
Our stance: We never say "you are angry" in workplace/education. We say "pace ↑ + visual hand on head + repeated clicks = confusion + low conviction → train, not fix". That is behaviour + intent + conviction scoring, not emotion label. That passes EU Act because we adapt delivery for empathy, we don't infer emotion to make decision about person.
EI = Emotional Intelligence. Not emotion detection.
It is empathy-based adaptive intelligence.
EI = Emotional Intelligence Layer — rebuilt as empathy-based adaptive intelligence, not Emotion AI. Patent 551315 evidence generated 2019, before generative avatars. Autonomous recruitment — candidate says "I'm confident" — transcript positive. Old bot passed. EI measured verbal flat + visual slouch → Conviction 42% LOW, Receptivity 78% HIGH → Engine 210 paused interview, simplified. Same camera, different outcome because we measured capability matrix and emotion levels for intent, not happy/sad label. This is the core distinction: Emotion AI labels, EI adapts.
Same 5 signals.
Emotion AI labels. EI adapts.
Book a 30-min evidence walkthrough.
Same pipeline — transcript + verbal + visual → scoring → live adaptation — but action changes per context.
EU banned labeling.
We never labeled. We adapted.
EU AI Act Art 5.1(f) 2024: Prohibits emotion inference in workplace and education institutions. If you say "this employee is angry/disengaged" and use it for performance decision — banned. Fine up to 7% global turnover.
EI passes because: We don't infer emotion to judge person. We measure conviction, attention, receptivity, intent pattern, behaviour under pressure to adapt OUR delivery — avatar tone, pace, simplify, suggest break, train. That is empathy-based adaptation, not emotion judgment. Legal opinion: falls under Art 50 transparency, not Art 5 prohibition, with CDR logging.
Board-auditable: Every adaptation traceable — Claim 7 (behaviour under pressure), Claim 8 (visual cue yawn=break), Claim 10 (attention scoring), Claims 1,15,17 (intent + ViGen). Zero retention — no face database stored. On-prem scoring possible. CDR F6,F7,F8 governs.
We started trying to help you finish your work.
That is the difference between Emotion AI and EI. One gave you a dashboard after the meeting saying you looked angry. Other gives you live help during the meeting — pause, simplify, train, suggest break — with human-like leadership, not module pusher. One was banned. Other is board-auditable.