Whitepapers
ALL UNGATED · PDFEngineering Intelligence: a constitutional architecture for AI over physical assets
The manifesto in formal paper form. Defines the category, states the three governing principles, specifies the seven-layer reference architecture and the six physics-provenance states.
Read paper · Download PDF →Engineering Library
The terms this platform relies on, defined precisely — with the governing standard where one exists and the page where each concept has its canonical home. The Library defines; it does not advise.
Browse the definitions →The Engineering Intelligence Ontology (EIO): entity and relationship specification
Canonical entity types, governed relationships, taxonomy alignment to ISO 14224, and the amendment process for ontology change. The specification is in drafting; its scope is published for comment until 2026-09-30.
Read the scope · Comment →Architecture notes
Physics validation as a non-bypassable gate
How constraint sets are assembled per asset class, evaluated, and recorded on failure — including the treatment of constraints that are themselves in dispute.
Residual publication in the digital twin fabric
Reconciliation method, tolerance definition, and why model retuning is treated as an engineering change rather than a maintenance task.
Unit safety and uncertainty propagation in ENGIN Solver
Dimensional typing of engineering quantities, parse-time failure on unit mismatch, and how measurement uncertainty reaches the displayed result.
Evidence citation in agent output
Node-level citation, document revision pinning, and the confidence model used across retrieval, diagnosis and planning agents.
Standards the platform is built against
Alignment is stated at the level of specific clauses in the architecture notes. Where the platform diverges from a standard, the divergence is documented rather than omitted.
Asset management system practice — decision records map to the standard's evidence expectations.
Reliability and maintenance data collection — the failure taxonomy in the EIO ontology.
Industrial security zones and conduits — the OT integration boundary is read-only by design.
AI management systems — governs how model output is labelled, reviewed and accepted.