Presence Mobility™ · NeoSoulTech Ltd

Human state
as a live control
variable.

An architecture-level shift. Not an AI feature. Not a wearable. Not autonomy.

Physical systems today are blind to the human operating them. They respond to mechanical inputs — not to the person behind them. Presence Mobility™ is the architecture designed to close that loop — embedding real-time human-state inference into deterministic control systems, with the human always in authority.

<50ms
Target inference cycle
Zero
Cloud dependency
Bounded
Co-regulation always
Physical
Presence
Systems
Human
The Shift

Neither wearables nor vehicles solve this. Both stop short.

Today — Wearables
Observe. Don't act.

Wearables surface human state data to dashboards, apps, clinicians. Excellent at observation. Disconnected from the physical system the human is operating.

Post-analysis, not real-time
Advisory only — human still decides manually
No integration with the control loop
Presence Mobility™
Infer. Modulate. Continuously.

Real-time physiological and biomechanical state inference, integrated directly into the physical control loop. Human variability becomes a live input — not noise to be ignored.

Target: inference in under 50ms, on-device
State shapes control delivery — never commands
Human remains sovereign — always, by design
Today — Vehicles
React. Don't know.

Vehicles are sophisticated mechanical responders. They optimise for torque, cadence, efficiency curves. The human behind the inputs is invisible to the system.

Physiologically blind
Same response regardless of rider condition
Mechanical optimisation only

"The objective is not autonomy. The objective is bounded co-regulation between human and machine. Every physical system that works alongside a human should know its human."

— Stevie Dymond, Founder · NeoSoulTech Ltd

System Architecture

Five layered domains. Safety-isolated by design, not by policy.

The AI inference layer is architecturally separated from the safety control layer. No inference output can breach the safety boundary under any condition.

01
Sensor Layer
What the system receives

Grip pressure, vibration signatures, saddle load, HR and HRV proxies, cadence, kinematic data, environmental and terrain inputs. Timestamped, confidence-scored, asynchronous.

Input
02
EPSV Engine
The inference core

Fuses multimodal inputs into a structured, continuously updated state representation — fatigue, stress proxies, stability, control consistency. Confidence-weighted. Degrades gracefully on signal loss.

Core
03
Intent Modulation
State → shaped execution

Translates inferred state into how control is executed — not what control is. Smooths torque under fatigue. Cannot exceed safety envelopes under any inference condition.

Bridge
04
Adaptive Control
The hard boundary

Regulatory limits enforced in dedicated RTOS firmware — speed, power, thermal, battery. This layer is architecturally unreachable by any AI software path.

Safety
05
Human Feedback
Minimal, coherent signals

Haptic, ambient, voice. Silence is the default state. Non-urgent advisories deferred to post-session. One-tap override returns to fixed-profile mode. Always.

Output
NeoSoul One

The intended proving ground. Not the proven product.

NeoSoul One is a modular electric mobility platform and the planned first physical embodiment of the Presence Mobility™ architecture. It is where we intend to prove the thesis — under the conditions that matter most.

Mobility is the right environment precisely because it is the hardest. The system must infer human state in real time, under motion, under physical consequence, within regulatory constraints. If the architecture holds here, the foundation for every subsequent domain is earned.

This is the work ahead. Not the work behind.

Prove: real-time inference under motion

Multimodal state inference must operate while the human moves — under vibration, environmental variation, and changing physical load. The lab is not enough.

Prove: safety architecture under consequence

The separation between AI inference and safety-critical control must hold when failure has physical cost. A boundary untested under consequence is not a validated boundary.

Prove: co-regulation within constraint

Bounded adaptation within regulatory limits — demonstrating that the architecture can shape behaviour without overriding it. The core claim of Presence Mobility™.

NeoSoul One · Architecture targets
Inference cycle target< 50ms
Sensor fusion target< 20ms
Hardware acquisition< 5ms
Compute power envelope< 5W avg
Grip sampling rate200Hz · multi-axis
Cloud dependency (safety)Zero — by design
Safety enforcementRTOS firmware · isolated
Biometric dataOn-device · user-owned
Regulatory scopeEPAC · US Class 1–3
Human overrideOne tap · always
ArchitectureVehicle-agnostic
PoC target2026
Scalability

Mobility is where we start. Not where we stop.

The Presence Mobility™ architecture is domain-agnostic. The roadmap moves from the most constrained environment outward — each step harder, each step building on the validation before it.

🚴
Proving ground
Personal Mobility

NeoSoul One. Inference under motion, safety under consequence, adaptation within regulation.

Next
Assistive Mobility

Adaptive mobility devices. Higher safety stakes, tighter operating envelopes.

🏥
Next
Rehabilitation

Recovery platforms where state-awareness must align with clinical safety constraints.

⚙️
Future
Industrial Systems

Human-operated machinery — fatigue, overload, degraded performance — within strict industrial boundaries.

🏭
Future
Factory & Production

Human-machine co-regulation at production scale, integrated into existing control architectures.

🏠
Future
Presence Home

Ambient presence intelligence. The least constrained environment — approached last, once the hard cases are proven.

Lower complexity
Higher complexity
Mobility is the most demanding validation environment available for this architecture — motion, physical consequence, regulatory constraint, real-time safety boundaries. We start here deliberately. Once the architecture holds here, the path to every subsequent domain is earned, not assumed.
Design Principles

Architecture-level commitments. Not configurable.

These are not product features. They are structural constraints on what the system is allowed to do and how it must behave.

01
Bounded co-regulation

All adaptive behaviour is bounded — cannot exceed predefined limits, operates separated from safety systems, degrades gracefully to baseline. Shapes. Never overrides.

02
Human sovereignty

One-tap override to fixed-profile mode at all times. Regulatory limits enforced in RTOS firmware — no AI inference output can breach them under any condition.

03
Edge first

All safety-critical intelligence executes on-device. Cloud connectivity is optional and consent-gated. The system must function fully without a network.

04
Privacy by design

Biometric data is user-owned, encrypted at rest, and never leaves the device without explicit per-category consent. Privacy is the default — not a setting.

05
Silence as default

The system's highest ambition is its own invisibility. Feedback only when it genuinely serves the human. When it is working correctly, you don't notice it.

06
Modular longevity

Compute, battery, and sensor modules are independently upgradeable. Architecture separability is structural. Platform life target: 10+ years.

The NeoSoulTech Ecosystem

One founding idea. Multiple expressions.

Presence Mobility™ is the primary platform. Alongside it, NeoSoulTech is developing physical AI presence at two additional scales — this is not a single-application thesis.

Primary Platform
Presence Mobility™
NeoSoul One · Edge AI · Personal Mobility

Human physiological state as a live control variable inside safety-bounded physical systems. NeoSoul One is the planned first embodiment — chosen deliberately as the hardest environment in which to validate the architecture.

EPSV / IMV architecture — real-time edge inference
Grip deconvolution — core patentable innovation
Vehicle-agnostic — designed to scale across categories and beyond
Safety isolation — AI inference separated from control layer by architecture
PoC target: 2026 · Patent filed: UKIPO
Also in development — Physical AI at the human scale
2026 · Computex Taiwan
AurenKeys
Physical AI Interface · Workspace Presence

A keyboard with an integrated Physical Presence Controller — hardware-level cognitive mode selection that operates AI as a silent cooperative layer during knowledge work. State broadcast to the Auren Orb via shared state layer. UK provisional patent filed.

2027
Auren Orb
Ambient AI Presence Device · Home & Workspace

A 90–120mm ambient presence object — frosted glass and ceramic, no visible seams or buttons — building a continuous model of its environment and cooperating with AurenKeys via a shared state layer. Calm technology: presence over performance.

The Team

Thirty years watching AI deployed away from people. Now building it back.

SD
Stevie Dymond
Founder & CEO · NeoSoulTech Ltd
30+ years · product development & manufacturing
Hardware · automation · IoT · AIoT
Product Dev Manager · Schneider Electric
Global Head of Quality · Sunsynk UK Ltd
Based in Taipei, Taiwan (Wenshan)
Company incorporated in the UK

Stevie Dymond brings 30+ years across product development, high-precision manufacturing, automation, IoT, and AIoT — primarily on factory floors in Taiwan and China. Prior roles include Product Development Manager at Schneider Electric and Global Head of Quality at Sunsynk UK Ltd.

The Presence Mobility™ thesis comes directly from that experience. Decades spent watching AI intelligence get deployed away from people — faster, more efficient, more remote — led to a single founding question: what would it look like to build the inverse? A physical system genuinely aware of its human, responding to that awareness in real time, never removing the human from control.

R&D and supply chain operations are based in Taipei — at the centre of the world's premier hardware and semiconductor manufacturing ecosystem, with direct access to Tier-1 manufacturing partners.

Company
NeoSoulTech Ltd · UK
R&D Base
Taipei, Taiwan
Stage
Pre-seed · Bootstrapped
Patent
Filed · UKIPO
PoC Target
2026
Address
Liverpool · UK
CompanyIncorporated · UK
PatentFiled · UKIPO
PoC DevelopmentIn Progress · 2026 target
AurenKeysComputex Taiwan · June 2026

Every system that works with a human should know its human.

We are looking for partners who understand what we are building and why it matters — investors, technology partners, manufacturing collaborators, and fellow builders.

NeoSoulTech Ltd · Level One, Basecamp · 49 Jamaica Street, Liverpool L1 0AH, United Kingdom
R&D Operations: Wenshan District, Taipei, Taiwan · www.neosoultech.com