SiteScope MultiModal™ Industrial Contextualization Platform
Industrial Data Context for Engineering and AI Workflows

SiteScope MultiModal is an industrial data contextualization blueprint. It connects raw time-series data, engineering documents, asset registries, event streams, work orders, simulation outputs, and domain ontologies into an operational knowledge graph that can be queried by people, applications, dashboards, and AI agents.

Its graph model covers more than oil and gas. SiteScope MultiModal targets discrete automation, process automation, buildings, energy systems, logistics, robotics, sustainability accounting, and carbon management through the same entity and relation model.

Ontology is the platform's context layer. Assets, tags, documents, alarms, workflows, behaviors, emissions factors, and business objects are represented as graph nodes and relations. Applications can visualize this graph or query it when preparing input for industrial AI workflows.

Core Platform
  • Industrial DataOps
  • Ontology graph
  • AI agent context API
  • Workflow activation
Design Principle
  • Deployment-independent services
  • Source-preserving ingestion
  • Versioned domain ontologies
  • Governed workflow interfaces
Reference Architecture

SiteScope MultiModal sits between industrial source systems and downstream workflows. It ingests source data, records semantic relationships, and exposes the resulting context through graph views, workflow interfaces, and APIs.

Layer Responsibility
Source connectorsConnect OPC UA, PLC tags, SCADA, DCS, MES, ERP, CMMS, BIM, CAD, historian, IoT, documents, spreadsheets, and carbon ledgers.
Contextualization engineResolve asset hierarchy, tag meaning, unit normalization, document links, event ownership, temporal alignment, and data quality.
Ontology graphRepresent domain concepts, asset relations, process topology, equipment capability, energy flows, emissions factors, and workflow state.
Graph visualizationLet engineers inspect ontology classes, entity instances, relationship confidence, lineage, and missing-context gaps.
AI context serviceProvide grounded context packages for agents, RAG pipelines, troubleshooting copilots, optimization agents, and planning workflows.
Activation layerExpose dashboards, alerts, field workflows, simulation loops, digital twins, and WFC orchestration.
Ontology and Graph Visualization Concept

The ontology graph is SiteScope MultiModal's core data model. It records how industrial data relates across source systems. A pump record can connect tags, P&ID symbols, work orders, failure modes, spare parts, energy consumption, alarms, 3D location, process function, and behavior capability.

  • Ontology authoring: domain experts define classes, relations, constraints, units, and naming rules.
  • Entity linking: tags, assets, documents, events, and workflows are matched into graph instances.
  • Semantic validation: missing relations, broken links, ambiguous tags, and low-confidence mappings are surfaced.
  • Graph exploration: users browse topology, lineage, root cause paths, and cross-domain dependencies.
  • Agent grounding: AI agents receive context bundles with graph paths, evidence, confidence, and governance rules.
  • Runtime activation: ontology relations drive alerts, recommendations, workflows, and simulation handoff.
Cross-Industry Context Model

The same contextualization pattern applies to process plants, automation cells, building systems, renewable assets, data centers, logistics networks, carbon accounting boundaries, and hybrid IT/OT operations.

DomainContextualized objectsAI agent use cases
Discrete automationRobots, CNCs, conveyors, PLC tags, quality events, work cells, takt-time signals.Throughput optimization, anomaly explanation, virtual commissioning support.
Process automationP&ID assets, loops, alarms, historian tags, maintenance records, batch events.Root-cause analysis, procedure recommendation, alarm rationalization.
BuildingsRooms, HVAC, meters, BMS points, occupancy, comfort zones, maintenance tickets.Energy optimization, fault detection, indoor environment reasoning.
Energy and gridSubstations, DER assets, power meters, weather, storage, dispatch events.Grid operation support, predictive maintenance, renewable forecasting context.
Dual-carbonEmission factors, activity data, product footprint, energy mix, carbon sinks, reporting boundary.Carbon traceability, abatement planning, compliance evidence generation.
Implementation Roadmap

Delivery starts by onboarding source data. Each later step adds explicit relations, validation rules, and provenance before the context is exposed to AI agents.

01 · Data Onboarding Connect

Connect historian, OPC UA, CMMS, MES, ERP, BMS, CAD/BIM, documents, simulation tools, and carbon data sources. Preserve source identity, timestamp, unit, document version, and access policy.

02 · Contextualization Relate

Build relations between assets, tags, documents, alarms, work orders, events, 3D objects, process topology, and carbon boundaries. Use rules, ontology constraints, embeddings, and expert validation together.

03 · Ontology Graph Model

Create a domain ontology layer for classes, properties, relations, constraints, lifecycle states, and behavior capability. Visualize the graph so missing context is visible before AI workflows depend on it.

04 · AI Context API Ground

Package graph neighborhoods, source evidence, confidence scores, time windows, and governance constraints into an agent input contract. The contract keeps each supplied fact tied to its source and operating context.

05 · Workflow Activation Act

Activate contextualized data through troubleshooting copilots, maintenance planning, energy optimization, emissions reporting, field work, WFC workflows, and SiteScope visual analytics.

AI Agent Input Contract

For industrial AI, raw data is rarely enough. SiteScope MultiModal prepares an input contract that is explicit, explainable, and governed.

  • Question intent and target domain.
  • Relevant graph neighborhood and asset topology.
  • Time-series windows and event sequence.
  • Linked documents, procedures, drawings, and work orders.
  • Units, constraints, confidence, lineage, and permissions.
  • Recommended action boundary and workflow handoff.
Intended Uses
  • Search: locate data, documents, tags, assets, and event paths through graph relations.
  • Explain: trace how an answer was assembled from graph evidence and source systems.
  • Operate: convert context into maintenance, optimization, energy, or carbon workflows.
  • Scale: reuse ontologies and contextualization patterns across sites and industries.
Copyleft© Ciel Z • CC BY-NC-SA 4.0