Explained
What is an Intuigent?
The short answer: a synthetic engineer. The longer answer is what that word has to mean before you should accept it from anyone, including us.
The definition
Each clause carries weight, so take them one at a time.
- Connects to existing systems. The P&ID set, the historian, the document store, the maintenance system. Nothing is replaced and nothing is migrated. The plant keeps its systems of record.
- Compiles a knowledge graph. Drawings, trends, and documents become one connected, queryable model of the plant: what connects to what, what protects what, which documents describe which asset. The compile takes under 48 hours. This is the step most industrial AI skips, and the reason most of it disappoints.
- Answers, then executes. It answers engineering questions in seconds, grounded in the graph. It also carries out the multi-step work that follows the answer: the trace, the evidence file, the draft deliverable.
- Traced to source. Every claim carries its citation: the sheet, the tag, the document revision. An answer you cannot check is not an engineering answer.
The distinction
What makes it an engineer, not a chatbot.
The difference is not fluency. It is four working habits that engineering demands and chat interfaces do not have.
It works from the compiled plant
A chatbot works from whatever documents fit in its window. An Intuigent works from the graph: the whole drawing set, the historian, and the document library, already resolved into one structure. Ask what sits upstream of an exchanger and it traces the connections, across sheets, the way an engineer would with the full set on the table.
It cites what it claims
Every statement in an output points back to a source: a P&ID sheet, a trend window, a procedure revision. The engineer reviewing the work can check any line of it without asking where it came from.
It refuses to guess
When the plant's data cannot support an answer, it says so, and says what evidence would close the gap. In an industrial setting, a confident wrong answer is worse than no answer. A system that cannot say “I cannot verify this” is not safe to put near engineering decisions.
It executes multi-step work with checkpoints
Isolation plans, evidence files for an investigation, scope packages: work with many steps and real consequences. The Intuigent plans the job, does the legwork, and pauses at the steps that matter for a human to review and sign. It does not proceed past a checkpoint on its own.
What an Intuigent is not
The word gets clearer when you draw the boundaries.
- Not a document search. Search finds the sheet that mentions a pump. It cannot tell you what the pump feeds, what isolates it, or what changed near it last quarter. Finding a document is the start of the work, not the work.
- Not a copilot bolted to a viewer. A viewer with a chat pane shows you the drawing and talks about the part you are looking at. The trace across twelve sheets is still yours to do by hand. Reading locally is precisely what a P&ID punishes.
- Not an autonomous system. No output changes anything in the plant without a person signing it. The checkpoints are architecture, not etiquette: the human owns judgment and accountability, permanently.
Why now
One in two process engineers retires this decade.
The judgment walks out the gate. The drawings and the data stay. Meanwhile the engineers who remain spend 40% of the day finding information the plant already has. A synthetic engineer that compiles the plant and does the legwork is not a fashion. It is arithmetic.