Cutting manual CAD work in a composites R&D team
A composites R&D team replaced slow, repetitive boundary and layup steps in CAD with assisted scripts, keeping the engineer in control of every decision.
R&D engineers design carbon-fibre parts in CAD. Certain boundary and layup steps are precise, repetitive and slow, and they vary subtly by part, so a single template breaks. The know-how sat with a few senior engineers, and the manual steps were a bottleneck on every new design.
Precise but repetitive
The boundary and layup steps ate senior engineers' time.
Every part is a little different
A naive template breaks, so the steps resisted simple automation.
An unforgiving platform
Small errors in the CAD automation cost real rework.
Knowledge with a few people
The method left the building when they did.
We built assisted scripts that handle the repetitive boundary and layup steps, parameterized to the part, with the engineer reviewing and adjusting each result. The scripts are advisory: they propose, the engineer disposes, and nothing changes without a check.
We shipped narrow and reliable first (the simplest geometry) and widened coverage only as each step was proven against the engineer's own test parts.
- Assisted boundary and layup scripts, parameterized per part
- Advisory output the engineer reviews in their existing CAD tool
- A tested baseline and a documented limit of where it applies
- No forced change to the existing workflow
Shipped narrow and proven, then widened
One expensive problem, proven before it scales. The A1 to POC method, run in weeks not quarters.
Senior engineers got their time back without giving up control.
- The repetitive boundary work drops from weeks to days.
- Senior engineers spend time on design, not on mechanical steps.
- The method is captured in tested scripts, surviving turnover.
- Advisory output keeps the engineer responsible for every decision.
We design and deploy to the ISO/IEC 27001 (information security) and ISO/IEC 42001 (AI management system) frameworks. Data stays where it should, decisions that carry real cost keep a human in the loop, and every model call is logged for audit.
Designed and deployed to these frameworks. Not a certification claim.
Have a problem like this?
Bring your messiest operational problem. In 30 minutes we will tell you where AI would pay back fastest, and where it would not.
Discuss an operations problem