Perception layer
Real-time pathSee the environment.
Sources of raw observation.
Multiple fixed cameras covering the same environment.
Moon 02 · Physical world
PWIP — Physical World Intelligence Platform
Give machines an understanding of the physical world.
PWIP is a reusable physical-world intelligence platform designed to understand people, objects, spaces, movement and events across camera-based environments.
Core concept
From pixels to understanding.
Every camera frame passes through the same path: observed, tracked, placed in space, checked against rules, and kept as evidence.
- 1
Cameras
Video from cameras already present in the environment.
- 2
Perception
Frames become structured observations.
- 3
Detection
People and objects are found in each frame.
- 4
Tracking
Detections are linked across time.
- 5
Identity / Re-identification
The same entity is recognised across views and gaps.
- 6
Spatial understanding
Observations are placed in zones and coordinates.
- 7
Rules + events
Rules evaluate the world model and create events.
- 8
Evidence + data
Snapshots, video references and logs back every event.
- 9
Intelligence
Dashboards, analytics and natural-language querying.
What PWIP understands
Eight questions about a physical space.
PWIP models an environment as people, objects, space, movement and time, then adds rules, events and evidence on top. Select a lens.
People — Who is present?
Individuals are detected, tracked and, where configured, recognised again across views.
Architecture
Seven layers, one world model.
Each layer has one job. Select the parts of a layer to see what they do.
Not in the real-time path
The LLM and intelligence layer is not in the real-time camera processing path. Detection, tracking, rules and evidence run without it; it works on stored events and structured data.
01 02 AI / vision layer
Real-time pathTurn frames into observations.
Computer vision models run on each stream.
Finds people and objects in each frame.
03 Tracking layer
Real-time pathFollow entities through time.
Lightweight, CPU-based tracking keeps the real-time path efficient.
Tracking that does not depend on a GPU per stream.
04 World model
Real-time pathOne shared picture of the environment.
Tracks from every camera are reconciled into a single model.
Each tracked person is one entity, however many cameras see them.
05 Rule engine
Real-time pathDecide what matters.
Rules evaluate the world model and create events.
Presence and absence over a period.
06 Evidence layer
Event-drivenKeep the proof.
Events are stored with the material needed to review them.
Still images captured at the moment of an event.
07 Intelligence layer
Outside the real-time pathAsk questions of the physical world.
The LLM and intelligence layer is not in the real-time camera processing path. It works on stored events and structured data.
Live and historical views of the environment.
Multi-camera understanding
One person. Four cameras. One entity.
As someone moves between camera views, PWIP represents them as a single tracked entity in a common world model, so a journey through a building is one continuous event, not four disconnected detections.
Common world model
One entity, 017, followed across four views.
- 14:02:11Enters at Gate ACamera 1
- 14:02:34Moves through corridorCamera 2
- 14:03:10Enters restricted areaCamera 3 · Rule matched
- 14:03:52LeavesCamera 4
Four cameras. Four detections. One continuous event, not four disconnected ones.
Rule engine
From video to events.
Rules describe what should and should not happen in each zone. When the world model and a rule disagree, PWIP creates an event, captures evidence, notifies an authorized user and stores the result.
IF person enters restricted zone
AND time > permitted period
THEN CREATE EVENT- 01Detect
- 02Evaluate
- 03Create event
- 04Capture evidence
- 05Notify authorized user
- 06Store event
Rules are evaluated against the world model, not against raw video.
Use cases
One platform, many environments.
A reusable foundation that is configured for each site: its cameras, its zones and its rules.
01Hospitals
- Staff attendance
- Movement monitoring
- Restricted zones
- Patient and staff flow
- Operational monitoring
02Government schools
- Student attendance
- Staff attendance
- Classroom presence
- Campus movement
- Operational monitoring
- Attendance and meal-programme correlation
03Factories
- Worker presence
- Restricted zones
- Safety events
- Production-area movement
- Operational intelligence
04Construction
- Site movement
- Restricted areas
- Worker presence
- Safety monitoring
- Event evidence
05Offices
- Attendance
- Room utilisation
- Restricted access
- Movement patterns
- Operational insights
A note on accuracy
Results depend on camera placement, lighting, coverage and configuration. PWIP is designed to support human review and decision-making, and is not presented as perfectly accurate.
Two systems
Two systems. One physical world.
HMS understands the hospital's information. PWIP understands the hospital's physical environment. Together, a physical event can arrive with its clinical and operational context.
HMS · digital world
Understands the hospital's information.
Patients, records, orders, results and services.
PWIP · physical world
Understands the hospital's physical environment.
People, objects, spaces, movement and events.
Physical world
People, spaces, equipment.
PWIP
Understands the physical environment.
Events / observations
Structured, with evidence.
HMS / operational systems
Understands the hospital's information.
Context
Physical events meet clinical and operational records.
Intelligence
Better questions, better answers.
Independent by design
HMS and PWIP are separate products. PWIP can operate without HMS, and HMS can operate without PWIP. Integration is optional, and neither is a feature of the other.
The universe
One company. Two moons.
AlienCore builds systems that connect intelligence to the real world. Each product stands on its own.
ALIENCORE
Optional integration
Start a mission
Understand the world your systems operate in.
Tell us about your environment, your cameras and the events that matter.