Manufacturing Copilot is available for commercial deployment. Discuss implementation
Contact

Manufacturing Copilot

Physical AI for plant operations

A unified knowledge layer. Predictive intelligence. Governed Physical AI. Built around your existing manufacturing systems.

Manufacturing Copilot

01/Foundations

Four foundations. One governed system.

Commissioned as a scoped implementation, with defined deliverables, acceptance criteria and support.

  1. 01

    Unified knowledge layer

    Connects engineering knowledge and operational data for contextual, source-grounded answers.

    • Manuals
    • SOPs
    • Maintenance records
    • Operational data
  2. 02

    Plant ontology

    Defines assets, process variables and relationships to establish consistent operational context.

    • Asset
    • Line
    • Sensor
    • Unit
    • Variable
    • Data source
    • Relationship
    • Limit
  3. 03

    Digital twin

    Represents plant state and supports predictive evaluation of proposed process changes.

    • Operating limit
  4. 04

    Independent control plane

    Governs AI authority, operating constraints and intervention approvals.

    • Within authority
    • Refer
    • Hard limits

02/Plant systems

Built around your existing manufacturing systems

Manufacturing Copilot connects operational intelligence across the ISA-95 functional hierarchy, from equipment and control systems to manufacturing operations and enterprise applications.

  1. L4

    • ERP
    • MRP
  2. L3

    • MES
    • Historian
    • Maintenance
  3. L2

    • SCADA / HMI
    • PLC
  4. L1

    • Sensors
    • IoT edge
  5. L0

    • Machines, lines and cells

Example interfacesOPC UAMQTT

Typical systems by level; placement varies by site. Interfaces are confirmed against your equipment and software versions.

03/Governed Physical AI

Intelligence proposes. Independent control governs.

Manufacturing Copilot separates AI reasoning from operational authority. An independent control plane evaluates and authorises interventions within defined operating limits, with human oversight where required.

  1. 01

  2. 02

  3. 03

  4. 04

  5. 05

Lower consequence

Within authority

Eligible actions may proceed automatically within approved authority and operating constraints.

Higher consequence

Refer

Goes to an authorised person, with supporting evidence.

Never crossed

Hard limits

Machine-specific operating constraints that no AI recommendation and no approval can override.

Technical detail: authority, evaluation and controlled execution

Bounded autonomy

  • A signed, versioned Agent Risk Passport defines what each agent is approved to do.
  • Runtime assessment weighs consequence, likelihood, uncertainty and current operating context.
  • Agents cannot change governance, protected ontology definitions or their own permissions.
  • Any expansion of authority needs risk assessment and human approval.

Intervention evaluation

Candidate interventions are evaluated on controlled plant-state snapshots with approved predictive models, including ML surrogate models linked to the digital twin.

Snapshot
Controlled plant-state snapshots.
Evaluate
Expected performance, reliability, uncertainty, model applicability and operating limits.
Coordinate
Shared resources, interventions already authorised and cumulative effects.

Controlled execution

Controlled execution
Changes reach the machine through an independently governed path; an agent holds no write access.
Bounded authority
Authorisation covers the target, parameters, conditions, sequence and validity period of the whole intervention.
Verified before execution
Asset identity, engineering units, control ownership and machine mode are checked immediately before execution.
Local control
Local controllers keep real-time control, and protective functions and interlocks stay independent of agent optimisation.
Runtime monitoring
Predicted and observed behaviour are compared, with a process-specific fallback if they diverge.
Everything recorded
Every intervention, approval, risk assessment, model version and outcome is recorded.

04/Delivery and deployment

Implemented by engineers who understand your plant

  • In productionUK-hosted AWS

  • Your cloudPrivate cloud

  • Your siteOn premises

  • Where supportedOffline or air-gapped

Forward deployed engineersIntegrate your systems, configure the agreed capabilities and commission the solution.

DevOps teamSets up hosting, security controls, monitoring and application delivery.

Data residency, inference, backups and logging are agreed within each deployment architecture.

05/Questions

Before you commission

What is Manufacturing Copilot?

Governed Physical AI for manufacturing and plant operations. Its Physical AI intelligence brings together a unified knowledge layer, plant ontology, digital twin and agentic AI; an independent control plane governs what that intelligence may do.

Can we commission Manufacturing Copilot directly?

Yes. It is commissioned as a commercially scoped implementation, with defined deliverables, commissioning, acceptance criteria and support.

Does it integrate with existing and legacy manufacturing systems?

Yes, through supported interfaces to selected systems across the ISA-95 levels, from PLC and SCADA to historians, MES and ERP. Compatibility is confirmed against your equipment and software versions before we commit.

What information can the unified knowledge layer connect?

Authorised manuals, SOPs, maintenance records and operational data. Responses identify supporting sources where available and flag missing or conflicting evidence.

Do we need autonomous machine control?

No. The knowledge layer, analytics and predictive capabilities run without autonomous equipment control; governed interventions can be added with engineering assurance.

How are AI decisions and physical interventions controlled?

Manufacturing Copilot has two layers. Its Physical AI intelligence, combining agentic AI, the digital twin and the plant ontology, analyses conditions and proposes interventions. Its independent control plane authorises, refers or blocks each one within the authority set by the Agent Risk Passport. Approved changes reach the machine through a controlled path, and local controllers and protective interlocks stay independent.

Where can Manufacturing Copilot be deployed?

It runs in production on UK-hosted AWS. Private-cloud, on-premises and supported air-gapped configurations are available, defined in the agreed implementation architecture.

Who implements and supports the solution?

Our forward deployed engineers handle integration, configuration and commissioning. Our DevOps team handles deployment and the agreed operational support.

What is the Manufacturing Copilot Cohort?

The Manufacturing Copilot Cohort is the collective group of manufacturers and businesses commercially deploying the product in their operations.

Commission Manufacturing Copilot for your plant