Physical AI Demo

When the physical world triggers an AI response.

See how DNotifier can turn a potential leak event into an intelligent investigation and governed operational response.

Oil & Gas Solution
What does this Physical AI demo show?

The scenario demonstrates how a Physical AI architecture can connect a gas or SCADA event to asset intelligence, specialized AI agents, physical inspection, decision-making and operational workflows.

PHYSICAL EVENT

DNOTIFIER EVENT MESH

EMERGENCY AGENT

ASSET BRAIN

MISSION AGENT

DRONE / CAMERA

LEAK AI

DECISION

HSE

CONTROL

WORKFLOW

The Scenario

One event. Multiple AI decisions.

The demonstration follows a hypothetical industrial incident involving an elevated gas reading near a pipeline. DNotifier connects the event with the asset, retrieves context, coordinates specialized agents and determines what evidence and response may be required.

01 · Detect

A physical signal changes

A gas sensor or operational system reports a reading that crosses a configured threshold.

02 · Investigate

AI asks what happened

An AI workflow retrieves the affected asset and relevant historical context.

03 · Respond

Physical evidence is collected

A configured inspection workflow can request a drone, camera or other physical inspection capability.

Autonomous Investigation Console

Watch the event move through the Physical AI stack.

This interface represents the type of event-driven operational workflow DNotifier is designed to orchestrate.

dnotifier://physical-ai/oil-gas/leak-investigation
Simulation
Event Stream
SENSOR_GAS_LEVEL_HIGH

Elevated gas reading detected.

PHYSICAL_EVENT_CREATED

Operational event created in DNotifier.

ASSET_CONTEXT_REQUESTED

Retrieve asset identity and history.

EMERGENCY_AGENT_ACTIVATED

Incident investigation workflow started.

INSPECTION_MISSION_REQUESTED

Physical verification requested.

DRONE_INSPECTION_COMPLETED

Inspection evidence received.

LEAK_ANALYSIS_COMPLETED

AI analysis completed.

RESPONSE_WORKFLOW_TRIGGERED

Configured response workflow initiated.

Asset Intelligence

PIPELINE-P-1024

StatusOperational
Asset TypePipeline Segment
LocationProduction Area 4
Previous FindingCorrosion · Medium
Last Inspection28 Aug 2026
Current Risk47 / 100

Evidence Available

Gas SensorAbnormal
TelemetryDeviation
Drone InspectionRequested
Thermal / OGIAvailable
AI DecisionInvestigate potential leak

The event is correlated with asset history and current evidence. Physical inspection is recommended before final incident classification.

End-to-End Workflow

From physical signal to operational action.

The important part is not any individual AI model. It is the orchestration connecting the physical world, AI reasoning and operational systems.

SensorInput
Event MeshEvent
Asset BrainContext
AI AgentReason
Drone / CameraAct
Inspection AIAnalyze
WorkflowRespond
Incident Timeline

What happens during the investigation?

01 · Gas event detected

A physical monitoring system produces an abnormal gas-related signal associated with a known operational area.

02 · DNotifier creates the event

The event enters the Physical Event Mesh and becomes available to the appropriate AI workflow.

03 · Asset context is retrieved

Asset Brain provides the affected asset's identity, location, history, previous findings and relevant evidence.

04 · Emergency Agent investigates

The Emergency Response Agent evaluates the situation and determines whether additional physical evidence is required.

05 · Mission is created

A Mission Agent can coordinate the required inspection workflow using an available drone, camera or other physical data source.

06 · AI analyzes inspection evidence

The Leak Detection Agent evaluates the collected evidence according to the configured inspection workflow.

07 · Risk is reassessed

New evidence can be incorporated into the asset context and used by the configured risk workflow.

08 · Operational response begins

Depending on the result and governance policy, the workflow can notify operators, HSE teams or trigger additional operational actions.

Multi-Agent Investigation

Different agents handle different responsibilities.

A complex industrial incident does not need to be handled by one giant AI agent. Specialized agents can collaborate through DNotifier's event-driven orchestration layer.

Agent 01

Emergency Agent

Coordinates the incident investigation and determines the next operational step.

Agent 02

Asset Risk Agent

Evaluates asset-specific evidence and risk context.

Agent 03

Mission Agent

Coordinates the physical inspection workflow.

Agent 04

Leak Detection Agent

Analyzes inspection evidence for potential leak conditions.

Decision Engine

The same event can lead to different actions.

Physical AI should not assume every alert requires the same response. The outcome depends on evidence, asset context, confidence, risk and configured operational policy.

Outcome A

False Alarm

Evidence does not support a confirmed incident. Record the investigation, update the asset context and continue monitoring according to policy.

Outcome B

Investigation Required

Evidence remains uncertain. Request another inspection, human review or additional sensor evidence.

Outcome C

Confirmed Incident

The configured response workflow can escalate the incident to responsible operators and downstream systems.

Operational Response

AI intelligence becomes useful when it reaches the right people and systems.

HSE

Route significant safety findings and incident information to the appropriate HSE workflow.

Control Room

Provide operators with incident status, evidence, asset context and AI recommendations.

Maintenance

Trigger the configured maintenance workflow when physical findings require intervention.

Enterprise Systems

Connect response workflows to appropriate enterprise systems through configured integrations.

Closed-Loop Intelligence

The investigation becomes part of the asset's memory.

A Physical AI workflow should not end when the incident is closed.

New inspection evidence, findings and operational outcomes can become part of the asset's historical context. Future investigations can therefore begin with more information about what has happened previously.

BEFORE PIPELINE-P-1024 │ ├── Previous Inspection ├── Telemetry └── Corrosion Finding ↓ PHYSICAL AI INVESTIGATION ↓ NEW EVIDENCE ↓ AFTER PIPELINE-P-1024 │ ├── Previous Inspection ├── Telemetry ├── Corrosion Finding ├── Leak Investigation ├── Drone Evidence ├── AI Finding └── Updated Risk
Developer View

Build the same workflow into your own application.

DNotifier is intended to provide infrastructure for developers building event-driven Physical AI applications, agents and operational workflows.

physical-ai-leak-workflow.js
const event = { type: "SENSOR_GAS_LEVEL_HIGH", assetId: "PIPELINE-P-1024", severity: "high" }; const asset = await dnotifier.asset.getContext( event.assetId ); const investigation = await dnotifier.agents.run( "emergency-response", { event, asset } ); if (investigation.requiresInspection) { await dnotifier.workflow.trigger( "physical-inspection", { assetId: event.assetId, mission: investigation.mission } ); } await dnotifier.events.publish( "INCIDENT_INVESTIGATION_STARTED", { assetId: event.assetId } );
What The Demo Demonstrates

Physical AI is an operational loop, not just an AI model.

Physical perception

Events can originate from sensors, cameras, drones and other physical-world systems.

Event-driven intelligence

Physical observations become events that can activate specialized AI workflows.

Persistent context

Asset Brain gives agents access to the history and operational context of the affected asset.

Multi-agent reasoning

Specialized agents can collaborate around detection, inspection, risk and response.

Physical action

Inspection capabilities such as drones or cameras can become tools within an operational workflow.

Closed-loop operations

New evidence can update the asset context and influence subsequent decisions and workflows.

Physical AI Demo FAQ

Questions about the DNotifier Physical AI demo.

Additional answers focused specifically on the autonomous investigation and response workflow demonstrated on this page.

The demo illustrates an event-driven Physical AI workflow in which a potential gas-related event triggers AI investigation, asset-context retrieval, physical inspection, evidence analysis, risk evaluation and a configured response.

See Physical AI In Action

Move from physical events to intelligent operations.

Explore how DNotifier connects physical data, AI agents, asset intelligence and operational workflows.