Aero turbine manufacturing with AI — the FAI is not rebuilt; it is generated while you build.

The problem

You make the part right; you rebuild the dossier every single time.

An aero turbine admits no excuses. You know how to make a good part: 5-axis machines, calibrated furnaces, NDT inspection by certified personnel, CMM in series. What hurts is not the part — it is the evidence that the part is good. Today that evidence rests on four systems that do not talk to each other:

1. The drawing and the specification — from the engine maker, revised through SR/SCN letters, living in PDM and in a parallel folder on the quality manager's drive.
2. Material traceability — the heat number of the melt, the foundry certificate, tied to each part lot through marks that manual OCR misreads on reflective superalloy surfaces.
3. The operation — the 5-axis recipe, the heat treatment furnace curve, the coating parameters, the NDT result. Each one on its own island.
4. The evidence pack — AS9102 FAI, NADCAP packages, the engine maker's own requirement. Assembled by hand, part by part, whenever it is asked for.

When the engine maker demands an FAI on a new part, the quality engineer disappears into a corner for a week to gather drawings, certificates, measurements, photos and signatures. When there is a complaint, months get reconstructed. The problem is not the part — it is that the evidence is built at the end, and everything built at the end weighs more than it should.

How it fits the IRIS system

iLEAN does not replace your PDM or your MES — it keeps the evidence pack alive between the two.

The FAI/NADCAP problem is not a lack of data: it is data on four different islands (PDM/PLM, MES, inspection systems, foundry certificates) that only get cross-referenced at the end. iLEAN acts as the putty that fills the gaps without asking you to change anything critical. And everything that touches AS9100, ITAR or EAR lives where it has to live — inside the three safety rings.

Edge sees every manufacturing and inspection operation. Connect ties traceability and certificate to the part. The agent assembles the FAI as you go — and the three rings guarantee ITAR/EAR. The person signs.

The three iLEAN pieces applied to aero turbine manufacturing:

• Edge — a machine-vision terminal (CNN) over critical stations: 5-axis output, NDT, coating, edge inspection. It detects surface defects, out-of-spec marks and NDT indications, and records the image tied to the part. It works with no network; what is critical does not wait for WiFi.
• Connect — captures the heat number by OCR even on reflective surfaces, ties the foundry certificate to the part, reads the furnace curve even when the controller is old, and captures the engine maker's SR/SCN changes through whatever channel they arrive on. And it also captures what comes from outside (NADCAP audit, complaint, regulatory change).
• Agent — knows the documentation tree of AS9100, AS9102 (FAI), the NADCAP special processes and the engine maker's own requirements. It assembles the evidence pack as the part goes through each operation. The quality engineer moves from building to reviewing and signing.

See the full IRIS architecture →

Before and after

Classic manufacturing vs. manufacturing with live iLEAN evidence

Aspect	Classic manufacturing	With iLEAN Edge + Connect + Agent
FAI / AS9102 on a new part	A week of a quality engineer's time	Generated as you go; the engineer reviews and signs
Heat number → part traceability	Manual OCR + Excel	Automatic reading + part tied to heat in line
Engine maker SR/SCN change	Arrives by email, distributed by hand	Connect captures it, the agent updates the live tree
Finding parts affected by an NCR heat	Manual reconstruction across systems	A direct query to the agent, in seconds
NADCAP pack (HT, NDT, coating)	Assembled for each audit	Live and traced by special process
Data sovereignty (ITAR / EAR)	Trust placed in heterogeneous silos	Three rings + cryptographic signature per plant

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Tier-1 or Tier-2 supplier to an engine maker, manufacturing blades / disks / casings in Inconel or titanium alloys, AS9100 and NADCAP in force, several programs with an active FAI.
• Edge + Connect + Agent pilot on one priority part family (the one that eats the most FAI time today). First value expected within a few weeks: automated heat number traceability and a first FAI generated as you go, ready for review.
• Indicative payback between 4 and 9 months, depending on the current volume of engineer-hours spent assembling FAIs and NADCAP packs, and on the cost of rework caused by incomplete or late documentation.
• A reasonable reduction to expect: ≥30% in engineer-hours spent on evidence, and a sharp drop in response time to a complaint or a NADCAP audit.

And the quality manager's reasonable doubt

“What if the AI builds the wrong heat number / part correlation and it slips into an FAI?” — hallucination is a problem of free generation, not of anchored tasks. Extracting a heat number from an image, reading a melt certificate and matching it to the part are anchored tasks; in those, the best models brought error below 1.5% ^[1]. And the quality standard in critical industries is measured in 25 PPM (parts per million) ^[2] — a level only reachable with systems that see everything and leave critical judgment to the person. What counts for certification is decided by the engineer, on top of a proposal signed by the corresponding ring.

^[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.

^[2] The 25 PPM quality standard in critical sectors — Symestic reference.