One engine, many contexts: it reads the scenario (never the person), crosses it with behavior and context, and answers with a suggested action that improves the experience first — media and offers only when they help. Pick a scenario and press Run simulation.
Proof of vision · All data simulated
1 · READ — the scenario, never the person (aggregated signals)2 · PERCEIVED — the scene in numbers (Sensory Box)3 · INTERPRETED — behavior science × the moment4 · SUGGESTED ACTION — experience first; media only when it helps5 · ANTICIPATED RECEPTION — hard news engineered before it lands
1 · What is READ — scenario & context signals fused in real time (always aggregated, never identity)
Journey — signals light up as the simulation runs
2 · What is PERCEIVED — Sensory Box (M3) turns the scene into numbers · detection never recognition
—
people in view
—
looking at screen (VAC)
—
avg attention dwell
—
reaction lift vs fixed loop
♀ women —♂ men —
"Reaction" = attention & dwell lift + redemptions — never facial emotion. Frames processed in volatile memory, discarded in milliseconds; only these numbers leave the box.
4 · The SUGGESTED ACTION — experience-first response on the right touchpoint (Trigger Yield M4)
QR
5 · ANTICIPATED RECEPTION — when hard news is coming, the landing is engineered before the announcement (Maister · framing · ESIM de-escalation · peak-end)
The moment coming
Unprepared reaction (predicted)
Prepared landing (what we do first)
Honesty is the method: the information is never hidden or softened into untruth — the environment, timing and tone are prepared so people receive it well and stay safe.
Live flow — READ → PERCEIVED → INTERPRETED → SUGGESTED ACTION → RESULT
Press ▶ Run simulation to see the engine at work. Each card is one real-time decision.
Redemption rate by trigger
Share of delivered offers redeemed · simulated
Journeys today, by hour
06:00 → 22:00 · simulated
3 · How it is INTERPRETED — behavior science crossed with the moment, every rule with a pedigree
A = replicated · B = solid, effect size varies · C = hypothesis our validation loop tests. No NLP, no facial-emotion AI, no debunked priming — ever.
Privacy by design: everything in this engine runs on aggregated, anonymized signals. Individuals are reached only through choices they declared themselves (opt-in interests, loyalty membership). Camera analytics = detection, never recognition — nothing stored. No political use, ever. LGPD-compliant by design, built to GDPR standards.