Technology

The technology that lets AI agents understand, analyze, and support engineering review

Not document or 3D-file search — we connect the identity, relationships, spatial context, and evidence of real engineering objects so AI can understand and analyze a project.

Process Rich Tag Graph

Pain point — Process data lives in SmartPlant P&ID or engineering lists, but AI cannot read shared object identity and relationships from raw files alone.

Taking the SmartPlant P&ID Database and Process/Equipment/Line/Valve/Instrument/Drawing Lists as inputs, we normalize object types, attributes, identifiers, and relationships against the DEXPI 2.0 specification, then generate a Process Rich Tag Graph. Without a P&ID, a list-based simplified graph is still possible, but connectivity and context are more limited than the full P&ID path.

What is this object, and what is it connected to?

Tag Resolver

Pain point — A document or drawing tag often does not map cleanly to one 3D object.

Tag Resolver resolves an engineering tag to the authoritative 3D target using identity, naming rules, hierarchy, and model context. Where the mapping is ambiguous it returns confidence, a candidate set, and evidence so humans can finish the decision.

Which real 3D object does this tag point to?

Spatial Graph

Pain point — 3D models hold geometry, but AI still needs structured object identity, hierarchy, relationships, and spatial context.

The Spatial Graph is primarily built from Smart3D — object identity, hierarchy, position, engineering relationships, metadata, and spatial context. Where needed, AVEVA RVM/ATT can supply an alternative engineering-model path, and Navisworks object properties / hierarchy can supply a simplified fallback. Richness is not equal: Smart3D is richest, then AVEVA RVM/ATT, then Navisworks properties/hierarchy.

Where is this object, and how does it relate to its surroundings?

Spatial Resolver

Pain point — Proximity search alone does not tell an engineer which neighborhood to review.

Spatial Resolver interprets engineering intent and document requirements against the Spatial Graph through qualified spatial rules. It selects the relevant neighborhood and relationships for design review — not nearest-neighbor search or scene-graph traversal.

Which space and relationships should be reviewed right now?

Evidence Snapshot

Pain point — Review judgments often leave no portable engineering evidence behind.

Evidence Snapshot packages the object, revision, Viewpoint, requirement or document clause, spatial rule or context, and review evidence used in a judgment. It is an evidence package for review — not a screenshot.

What did the AI look at, and why did it reach that judgment?

File search → Object identity → Relationships → Spatial context → Analysis → Design review → Evidence

AI analyzes, identifies candidate issues, and supports design review. Final approval and disposition stay with the people who hold authority.