Gottlieb, erklärt in drei Minuten.

Gottlieb. Engineering AI for the built world.

Before a part is released for series production, your team or a supplier prepares up to forty-three document types, across eight disciplines. From design specifications, to functional safety, and all the way to the production process. Every one of them has to agree with every other one. Three are missing here. And every one of them is dense. A detail drawing: every dimension and tolerance, and one note that makes a radius critical. Seven sheets. A design FMEA: failure modes, causes and controls, nested three levels deep. Twenty-eight sheets. A test report: four hundred measurements against the limit line. Forty-four pages. Every change has to reach every document. That is where gaps can hide.

Here is how we check it. Code pulls out every value. AI reads what code cannot: drawings, scans, prose. One hundred and eight agents judge the facts, across eight lifecycle gates. Code links every fact into one graph. And cross-checks it against ten thousand written rules, in five layers. Director agents set the verdict. A separate check audits it. Your engineer decides. Same package, same answer, every time.

Now the hard part. It starts with the drawing. AI reads it into facts, and the facts become a node. The risk analysis and the fatigue test, read the same way. Three nodes, one graph. The drawing changed in May. The risk analysis still rates it low risk, because the fatigue test passed. Check any two documents against each other, and every pair passes. Walk the chain, and the rating rests on a test of samples built before the change. A geometry that no longer exists. That gap lives on the chain, not on any pair. The graph finds it, with three page numbers.

The result: not approved. Fourteen findings, five critical, three documents missing, and the wrong submission level. Every finding cited to its page, in five reports: for the supplier, the engineer, the auditor, and the board. The engine prepares. The engineer decides.

Next: every package becomes a knowledge graph you can ask. Programmes, plants, suppliers, parts, requirements. Nobody maintains it. It is built from the evidence you just saw checked. Now ask it the questions that matter.

A change comes in on the motor. What does it touch? Nine configurations on a second programme, four carryover modules, three plants. Known in minutes, before the change is released.

Which carryover modules entered the new architecture on an approval written for the old one? Four of six. The parts did not change. Their validity did. Re-qualification computed from what changed underneath.

The same gap costs ten times more at every later stage. Found before the start of production, the part and its production process can be adjusted.

Gottlieb. Engineering AI for the built world.