Medical and ethics advisor (MD, PhD)
Medical and ethical advisory regarding interpretation boundaries, clinical communication and responsible deployment of physiological indicators.
Under the Hood
You've seen the outputs. Now see the architecture behind them.
Five layers. Patented. One continuous loop that follows a person's state over time, running on the device by default.
A system that already sees its surroundings still misses the person. This page describes, in general terms, how EmoPulse adds that layer. Details are shared with partners on request.
Section I
How it works, in short
EmoPulse reads signals a system already has, such as the camera and the microphone, and turns them into a small, structured picture of the person's state. The picture is updated continuously and compared with the person's own normal, not with an average. The result is handed to the system as a signal it can act on.
- 1
Layer I · Signals
Read what is already there
Camera, voice and text streams are read side by side.
- 2
Layer II · Parameters
Measurable, not labelled
Raw observations become measurable parameters, not emotion labels.
- 3
Layer III · Coherence
Streams checked against each other
Where they agree and where they do not is itself information.
- 4
Layer IV · State
One compact picture
Everything is assembled into a single machine-readable state.
- 5
Layer V · Time
Against the person's own baseline
The state is followed over time, so change is visible and not only the moment.
Technical detail is shared with partners on request: info@emopulse.app.
Section II
The Machine Cannot See The Room
Most systems today operate from descriptions of people rather than direct observations of their current state.
The architecture exists to close that gap.
The Argument
FOUNDER"Current systems operate from predefined assumptions about the user. This architecture operates from observed state."
FOUNDER"The important thing is not the demonstration. The important thing is the architecture. A system that receives structured information about human state no longer has to rely entirely on assumptions, questionnaires or predefined rules. It can respond to what is actually observed."
What That Means In Practice
A pilot reports that everything is normal while physiological indicators suggest increasing fatigue.
An operator declares readiness at the end of a fourteen-hour shift because that is what professional operators do.
A patient completes a questionnaire according to how they believe they should answer rather than how they currently feel.
None of these people are necessarily misleading anyone.
They are describing themselves.
A description is not the same thing as a measured state.
Today, most systems rely primarily on text, forms, questionnaires and self-report.
Meanwhile, behavioural and physiological signals remain present in devices that already exist, but are not interpreted.
Our objective is simple:
To compute those signals locally, on-device, without requiring images or biometric streams to leave the device.
Modern AI systems possess instructions, search capability and reasoning capability.
What they generally do not possess is a real-time human-state input.
Section III
The Builder
Founder & CEO · Sole Inventor · ProBuggy #5
Before EmoPulse, my life was measured in dust clouds, broken components and high-speed decision-making.
I competed professionally in off-road racing series including LOORRS and the AMSOIL Championship, driving ProBuggy #5.
People often assume racing is about speed.
It is not.
The difficult part is maintaining structure under uncertainty.
A small mistake can end a race, destroy equipment or cause injury.
Racing teaches pattern recognition under pressure.
You learn to detect anomalies through vibration, sound, behaviour and environmental cues. You learn that noise is not always interference. Sometimes noise contains the earliest indication that something important is changing.
Years later, while building EmoPulse, I recognised the same principles inside AI systems.
Pattern recognition. Signal extraction. Human-in-the-loop discipline.
Systems rarely fail because they are too slow.
They fail because context is lost, meaning drifts or noise overwhelms interpretation.
The objective of EmoPulse became understanding that process in a structured, computable way.
EmoPulse was not born from a theoretical exercise.
It emerged from environments where decisions carried consequences and where structure mattered.
Whether on a racetrack or inside an AI system, the same principle applies:
If structure is ignored, reality eventually enforces it.
- RACING
- LOORRS · AMSOIL Championship · ProBuggy #5
- PATENTS
- Patented architecture
- SYSTEM
- Human-state layer · On-device by default
- STATUS
- Public proof-of-concept demonstrators · Internal measurements only · No independent evaluation yet · Optional enterprise server mode · A joint proposal submitted with partners in several countries
Advisory
Why It Runs On The Device
Many regulated sectors increasingly treat transmitted biometric information as a liability rather than a feature.
The EU AI Act places restrictions on certain forms of emotion inference in sensitive environments.
Healthcare, defence and security programmes frequently require stronger control of biometric processing and data movement.
Our response is architectural rather than contractual.
In the strict deployment model, computation remains entirely on the device.
There is no server-side biometric copy to leak, attack or retain.
The system continues operating when network connectivity is degraded or unavailable because connectivity is not required for operation.
Where real-time sharing is necessary, for example during a clinical consultation, state information may be transmitted under the rules of that deployment, including consent, isolation and audit controls.
The operating regime is therefore a property of the architecture rather than a promise made about it.
It can be verified directly.
Open the network panel in the Signature demonstrator and inspect the traffic yourself.
Every system that already sees the environment will need the person in it.