Anenergyontologyyouragentsandyourplannerscanbothread
Grid operators do not lack data. They lack one place where an asset, its condition, its constraints, its work history and its regulatory obligations are the same object.
Fromeightsystemstooneobjectmodel
A transformer, its substation, its work orders, its inspections and the agent authorised to act on them — typed objects and named relationships that a planner and an agent both read.
- Engineer—ASSIGNED_TO→Work order
- Maintenance agent—AUTHORIZED_FOR→Work order
- Work order—RELATES_TO→Transformer
- Transformer—LOCATED_AT→Substation
- Inspection—INSPECTS→Transformer
- Inspection—PRODUCES→Evidence
- Grid incident—AFFECTS→Transformer
Whythissectoristhehardestandthemostrewarding
A distribution or transmission operator runs an asset base measured in decades, a control system that cannot be disturbed, an inspection regime with legal weight, and a planning function under pressure from electrification and connection queues. The information required to make a good decision about a single asset is typically spread across a geographic information system, an asset management system, a historian, a pile of inspection reports and the memory of an engineer who has been there twenty years.
That fragmentation is why grid decisions take as long as they do — not because anyone lacks judgement, but because assembling the context consumes most of the available time. It is also why generic AI tools disappoint here: a model that can summarise an inspection report still cannot tell you which other assets share the failure mode, which work orders are already scheduled against them, or whether the constraint you are managing makes the intervention safe this week.
An ontology fixes the assembly problem. Once an asset, its location in the network, its condition history, its scheduled work, its constraints and its obligations are one object, both a planner and an agent can reason about it.
Theenergyontology
The object types that recur across network operators, and the relationships that make them useful.
Assets
Transformers, cables, switchgear, substations and meters, with condition, age and criticality.
Network topology
How assets connect, so an impact question can be answered by traversal rather than by a phone call.
Work orders
Planned and executed interventions, their windows, crews and dependencies on asset state.
Constraints
Capacity and operational limits, and which assets and customers they bear on.
Incidents
Outages and deviations, linked to affected assets, customers and root cause.
Inspections
Findings over time, typed against the asset rather than filed as documents.
Risks
Assessed failure likelihood and consequence, connected to the assets that carry it.
Connections
Requests and agreements, with their dependency on available capacity.
Obligations
Regulatory and licence requirements attached to the assets and processes they govern.
Decisionsthatgetbetter
Each of these is decision support: the agent assembles and proposes, a qualified person decides.
Maintenance prioritisation
Ranking intervention candidates by condition, criticality, network consequence and access window — with the reasoning attached rather than a score alone.
Inspection triage
Turning inspection findings and imagery into typed observations against assets, surfacing shared failure modes across the estate.
Outage and work planning
Assembling the constraints, dependencies and customer impact of a proposed window before it goes to planning.
Congestion and constraint analysis
Explaining which assets and connections a constraint affects, and what the options are — as analysis into the existing operational process.
Connection queue assessment
Bringing capacity, topology and planned reinforcement together so a connection enquiry can be answered consistently.
Engineering knowledge access
Making standards, procedures and prior decisions findable with citations, so precedent is available rather than remembered.
Field decision support
Giving a crew the assembled asset history, applicable procedure and safety constraints on site.
Regulatory reporting
Assembling reports from the same objects the operational decisions were made against, with the evidence intact.
Theboundary,drawnexplicitly
This is the part of an energy conversation that has to be unambiguous.
OT zone
Process control. Read paths only, through a broker.
IT zone
Where the ontology, the control plane and the agents live.
What NeuroCluster does not do
It does not perform autonomous grid control. Anything that changes the state of the network goes to a qualified operator with the case assembled, inside the systems and procedures already approved for that purpose.
Questions
- Does anything connect directly to our SCADA system?
- Only through a read-only broker, rate-limited and logged, and usually via a historian rather than the control system itself. No write path exists from the platform into process control.
- How long before an ontology is useful?
- Useful comes from modelling one decision, not the whole estate — typically the asset, its topology, its work history and the constraint that bears on it. Enterprise-wide modelling before a first workload reliably produces something nobody uses.
- Can this run without sending data outside our environment?
- Yes. On-premises and disconnected deployment with local inference on open-weight or private models is supported, which is a common requirement in this sector.
- How does this relate to our existing asset management system?
- It sits over it, not instead of it. The asset management system stays the system of record; the ontology defines what its objects mean in relation to topology, constraints, inspections and obligations held elsewhere.
Continue
- Critical infrastructureThe cross-sector IT/OT pattern in full.
- Energy-aware computeCompute as a grid-aware load.
- OntologyThe mechanics of the semantic layer.
- AgentsMandates, approvals and evidence.
- Edge AIDecision support at distributed assets.
- Use casesSpecific workflows with their evidence level stated.
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