Computable Contextual Knowledge Graph for Industrial Intelligence

The Contextual Knowledge Graph is a multi-layer computable model of your industrial facility, built by reading your existing engineering documents and operational data. It extracts the engineering knowledge from your P&IDs, isometrics, data sheets, process narratives, and HazOp reports, then maps the relationships between them — connecting every piece of equipment, every instrument, every pipe, and every safety barrier across all document types into a single, queryable model. It is the foundation that powers all three Drishya products: Artisan for engineering digitalization, Brains for predictive operations, and Junior for natural language queries.

The Graph reads P&IDs (Piping & Instrumentation Diagrams), isometric drawings, process flow diagrams (PFDs), heat an mass balance sheets, mechanical and PSV data sheets, process and control narratives, HazOp reports, corrosion control documents, shutdown key documents, single line diagrams (SLDs), and bills of materials. Input formats include PDF, JPG, JPEG, PNG, and CAD (DWG). For operational data, it connects to historian systems (OSIsoft PI, IP.21), DCS, SCADA, and PLC systems. If your facility has the document, the Graph can read it.

Most digital twins are 3D visual models of a facility — useful for spatial navigation but limited in engineering intelligence. The Contextual Knowledge Graph is a computable engineering model. It doesn't just show you what the facility looks like. It understands how it works — the process connections between equipment, the control logic governing operations, the safety barriers protecting each system, and the operating conditions each component was designed for. A 3D digital twin tells you where a valve is located. The Contextual Graph tells you what that valve controls, what stream flows through it, what its design limits are, what safety system protects it, and what its current sensor reading says.

A document management system (like AVEVA, Hexagon SDx, or Aconex) stores, organizes, and retrieves your engineering documents. The Contextual Knowledge Graph reads them. A document management system knows you have a P&ID named "PID-901-12-002" filed under Project X. The Graph knows that P&ID contains Valve XV-101, which connects to Heat Exchanger E-201, which is monitored by Temperature Transmitter TT-1204, which has a design limit of 350°C. The difference is between filing and comprehension.

Brownfield facilities are exactly what the Graph was built for. Over 95% of industrial plants are more than 10 years old, have no digital model, and rely on engineering documents that are decades old — often in inconsistent formats, mixed drawing standards, and legacy tag naming conventions. The Graph handles all of this. It reads legacy PDFs regardless of age, format, or drawing standard. It reconciles inconsistencies across document sets. And it builds a computable model from whatever your facility has today — not from what an ideal digital-first facility would have. No new sensors. No re-drawing. No clean data require.

The Graph is additive. Every new document you upload — a revised P&ID, an additional ISO, an updated data sheet — is read, processed, and integrated into the existing model. Relationships are automatically updated. New entities are connected to existing ones. Revision histories are tracked. When you connect operational data through Brains, the Time-Series Layer activates, adding live sensor readings and alarm data to the engineering context that already exists. The model starts with whatever documents you have today and deepens with every document and data source you add — growing from a partial facility view into a complete, living intelligence layer.