Walkaround

Today in aviation

Technical and maintenance

GE and Pratt put AI in front of the borescope

GE Aerospace is extending its AI blade inspection tool from the GEnx to the CFM LEAP and GE9X, reporting a borescope inspection cut from three hours to ninety minutes. Pratt and Whitney bought an AI borescope company this year.

15 October 2026

GE Aerospace has expanded its AI enabled Blade Inspection Tool beyond the GEnx to the CFM LEAP and the GE9X. The tool reviews borescope imagery and tells the technician which frames are worth a closer look, rather than replacing the judgement about what is in them. GE reports inspection process time halved, from around three hours to ninety minutes, and says accuracy improved against a standard borescope inspection. It has been running on the GEnx for about three years and is deployed across more than a dozen of the company's own shops.

Pratt and Whitney has moved the same way. It acquired the Dutch company Aiir Innovations this year, whose software analyses borescope footage to flag potential findings more consistently. That technology has already been in use with commercial operators and MRO providers.

Aviation Week's own assessment this month is worth holding alongside the announcements: dozens of companies are now selling AI tools into maintenance, and the market risks being overloaded with options.

In plain terms

A borescope is a camera on a flexible probe, pushed into an engine through an access port so the inside can be inspected without taking it apart. The difficulty has never been getting the picture. It is deciding what a mark on a blade actually is.

On the line

The reported gain is time, but the interesting claim is consistency. Borescope interpretation is one of the most experience dependent calls in the shop, and two competent inspectors can read the same frame differently. A tool that narrows that spread is worth more than one that only runs faster, because variability is harder to plan around than duration. Judge an implementation on what it writes into your maintenance system rather than on the model: a finding that arrives early enough to become a planned removal is worth a great deal, and the same finding surfaced in a portal nobody opens is worth nothing.

Sources