Theengineeringknowledgethatcurrentlylivesinpeople
Industrial operations run on expertise that was never written down. The most valuable thing AI can do here is make that expertise available without pretending to replace it.
Theplantontology
A machine, the line it belongs to, the work order against it, the technician assigned, the quality check performed and the downtime event that started it all — connected, typed and traversable.
- Technician—ASSIGNED_TO→Work order
- Maintenance agent—AUTHORIZED_FOR→Work order
- Work order—RELATES_TO→Machine
- Machine—PART_OF→Production line
- Quality check—INSPECTS→Machine
- Quality check—PRODUCES→Evidence
- Downtime event—AFFECTS→Machine
Whereitapplies
Asset reliability
Condition, failure history and criticality as one object, so prioritisation is explainable.
Maintenance
Intervention planning with the constraints, spares and access windows assembled.
Quality
Deviation investigation across process data, inspection results and batch records.
Inspection
Structuring findings and imagery into typed observations against equipment.
Engineering knowledge
Standards, drawings, procedures and prior decisions, findable with citations.
Digital twin context
Supplying the semantic layer a twin needs to connect to business objects and obligations.
OT integration
Read-only telemetry through a broker, respecting existing segmentation.
Supply chain
Supplier, part and lead-time context connected to the assets and orders that depend on it.
Operational intelligence
Cross-site comparison on a common model rather than on differently-built reports.
Ondigitaltwins
Most industrial organisations have or are building a twin of some kind, and most twins are strong on physics and weak on context. They can tell you how a machine is behaving; they cannot tell you which contract obligation is at risk, which crew is qualified, or what was decided the last three times this happened.
An enterprise ontology is the complementary half. It connects the physical model to the business objects — contracts, work orders, obligations, decisions — that determine what the physical behaviour means commercially. The two together support a decision; either one alone supports an observation.
Nothing in this pattern writes to process control. Telemetry is read through a broker; proposals go to a qualified engineer inside the existing operational system.
Questions
- Can this work with no connectivity on the plant floor?
- The edge runtime addresses exactly this and is in development. Central deployment with a local read broker is available today, which covers most inspection, quality and maintenance work.
- We already have a historian and a CMMS. What is added?
- The relationships between them, expressed as one model, plus a governed way for agents to reason over that model and leave evidence. Neither system is replaced.
- How do we capture undocumented expertise?
- Mostly by capturing decisions. When experienced engineers make and record judgements against typed objects, the reasoning accumulates as evidence — which is a far more reliable route than trying to interview knowledge out of people.
Continue
- Edge AIInference at the plant.
- Critical infrastructureIT/OT segmentation in full.
- OntologyConnecting physical models to business objects.
- KnowledgeDrawings, standards and procedures with citations.
- IntegrationsHistorians, CMMS and the broker pattern.
- Energy & utilitiesThe adjacent asset-intensive pattern.
Bring us one operational problem.
You do not need a finished brief. Bring the problem — we will work out the next step together.
Or book a call with the team