Engineering System of Truth
ESOT is the authoritative record of what an asset is, what it was designed to do, and what has been decided about it. It reconciles the design basis, the as-built condition, the maintenance history and the live operating state into one version that engineering will stand behind.
- 01Ingests P&IDs, datasheets, FMEA and RCM studies, CMMS history and historian tags
- 02Resolves conflicting sources through explicit precedence rules, not silent overwrite
- 03Every field carries its source document, revision and date of last verification
Engineering Intelligence Graph
The EIG holds assets, systems, components, failure modes, documents, measurements and decisions as typed nodes with typed relationships, all resolved against the EIO ontology. It is what makes questions like "which decisions depend on this datasheet revision?" answerable in one hop instead of one week.
Engineering Digital Twin Fabric
A twin is only useful if it disagrees with reality out loud. EDTF continuously reconciles the physics model against live measurement and publishes the residual — the gap between what the asset should be doing and what it is doing — as a first-class engineering signal.
When residuals drift outside tolerance, the fabric does not silently retune the model. It raises a reconciliation exception for an engineer, because a growing gap is usually the asset degrading, not the maths being wrong.
Deterministic computation, so the answer is the same every time you ask.
The solver evaluates governing equations — mass and energy balance, thermodynamic state, pump and fan affinity, heat exchange effectiveness, fatigue and remaining-life models — against the asset record. Results are reproducible, unit-checked, and returned with the equation and inputs that produced them.
Unit-safe by construction
Every quantity carries dimensions. A computation that mixes bar and kPa fails at parse time rather than producing a plausible wrong number.
Traceable to the equation
Each result links to the governing relation, the reference it came from, and the specific input values — the same evidence an engineer would attach to a hand calculation.
Tolerance-aware
Measurement uncertainty propagates through the calculation, so a result near a limit is reported as near a limit, not as a false precision.
Physics validation as a gate, not a score
Every proposed action — a setpoint change, a deferred inspection, an optimisation — is evaluated against the asset's constraint set before it is displayed. Actions that breach a conservation law, a material limit or a certified envelope never reach the engineer as a suggestion.
Blocked actions are not hidden. They are recorded with the specific constraint that failed and the margin by which it failed, so the reasoning is inspectable and arguable.
One agent per asset, retrieving and proposing under supervision.
Each asset carries its own agent, holding that asset’s design basis, history and open exceptions. Agents operate over the graph, not over raw documents. They cite the nodes they used, state a confidence value, and hand every conclusion to a named engineer for acceptance. No agent writes to the record of truth on its own authority.
Knowledge agent
Answers engineering questions from the graph with citations to the exact document revision and clause.
Failure-mode agent
Ranks candidate failure modes against observed symptoms and the asset's own FMEA, never a generic library.
Intervention agent
Proposes inspection and maintenance timing, with every option passed through PhysicsNET first.
Compliance agent
Checks that decisions and their evidence satisfy the standards and internal procedures in force for that asset class.
The output is an auditable engineering decision.
Not a chart, not a chat transcript. A versioned record naming the asset, the question, the evidence, the physics result, the AI contribution and confidence, the engineer who accepted it, and the date. It survives an audit and it survives staff turnover.
Deployed under the same discipline as the plant it observes.
Segmented by design
Read-only OT integration through a unidirectional boundary. ENGINPILOT observes; it does not write to control systems.
Standards alignment
Architecture mapped to IEC 62443 zones and conduits; asset management practice aligned to ISO 55000; failure data to ISO 14224.
Data residency and deployment
Available as a single-tenant private deployment or in-region managed service. Engineering documents never leave the customer boundary without explicit configuration.