AI should not only understand the world. It should act in it — safely.
IoT DC3 is an open-source Physical AI Runtime: the execution layer between an agent’s intent and the physical world.
Models reason. IoT DC3 senses, connects, governs and executes.
- 28
- Protocol drivers
- 330+
- Agent tools
- OAuth 2.1
- Secure MCP
- 10 years
- Industrial IoT engineering
The Physical AI Loop
Two worlds, one runtime
Devices on the left, intelligence on the right, IoT DC3 in between.
Intelligence is leaving the screen. Infrastructure has to catch up.
In software, a wrong action can often be retried. In the physical world, every action changes state, carries risk and needs accountability.
The world must be observable
Devices, telemetry, alarms and operating context need a stable semantic surface.
Execution must be governed
Identity, permissions, whitelists and risk tiers must sit between intent and action.
Effects must be traceable
Every command needs an auditable path from intent to execution and feedback.
A model output is not yet an industrial action.
IoT DC3 sits between agent intelligence and physical execution
It does not replace the model or business policy. It provides the context, tools and safety boundary they need to reach the field.
AI Agent
Understands goals, reasons with knowledge and decides what should happen.
ReasonPlanDecide
IoT DC3
Structures context, exposes tools, authorizes calls, executes commands and returns feedback.
ContextTool GatewayPolicyAudit
Physical World
Devices produce state. Commands create effects. The environment closes the loop.
DevicePointCommand
Five capabilities turn reasoning into physical action
The loop is complete only when an agent can see, understand, decide, act and verify the effect.
Sense the field
Read devices, telemetry and alarms.
Structure context
Turn physical signals into model-ready context.
Reason with knowledge
Combine models with industrial rules and history.
Execute safely
Call tools and issue commands through policy gates.
Verify every effect
Observe outcomes and preserve a full audit trail.
Two principles keep the surface simple and the execution trustworthy
One principle defines how capabilities are exposed. The other defines how actions are governed.
Everything is a Tool.
A unified surfaceDevices, points, commands and APIs become tools with explicit schemas. Agents use one consistent contract instead of learning every protocol.
DevicePointCommandAPI
Every Action is Traceable.
A governed execution pathIdentity, authorization, risk, execution, feedback and audit form one chain. Safety is part of the runtime, not an afterthought.
AuthRiskExecuteFeedbackAudit
Responsibility is separated so the system stays controllable
A reliable Physical AI system is not one all-powerful agent. It is a collaboration with explicit ownership.
Decides what should happen
Goal understanding, reasoning, planning and business judgment.
Governs how it may happen
Context, tool contracts, authorization, execution, feedback and audit.
Produces the physical effect
Real state changes under field constraints and safety mechanisms.
Intelligence can evolve quickly because the execution boundary remains stable.
The foundation remains. The role moves forward.
Ten years of industrial IoT engineering is not the old story we leave behind. It is the prerequisite for the new one.
- Foundation
Open-source IoT platform
Protocol access, device models, cloud-native architecture and industrial reliability.
- Expansion
AI-powered industrial IoT
Spring AI, MCP and agent capabilities become native platform building blocks.
- Direction
Open-source Physical AI Runtime
A safe and observable execution layer connecting agents to the physical world.
Explore the project from the path that fits you
Read the technical documentation, follow the complete engineering story, or experience an industrial dashboard directly.
Build the Loop
Bring your agents into the physical world — without losing control.
Start with open-source code, connect a device and turn one intent into one traceable physical action.
LIVE INDUSTRIAL DATA FLOW
Keep devices, data and intelligence in motion
From heterogeneous device access to AI-driven decisions, IoT DC3 turns every data change into a visible and controllable real-time loop.