I-care buys DiagFit to flag faults on satellites and helicopters that rarely fail

A reliability engineer monitoring a helicopter gearbox usually has years of vibration data and almost no failures in it. That asset is the customer for what I-care bought. DiagFit uses historical operating data to establish how a complex asset behaves under normal conditions, then compares incoming data against that baseline ^[3], and it does not need a large library of previous failures to identify an anomaly ^[4]. Amiral's technology has already been applied to satellites and helicopters ^[7], the kind of assets that accumulate extensive operational data without producing enough failure events to build a conventional failure-based model ^[8].

What lands on the planner's screen is a deviation. The output is an indication that something has changed ^[3]. A technician still has to decide whether to open the casing. I-care did not disclose what it paid, and the announcement puts no number on how often a flagged deviation turns out to be a real fault ^[13].

What the deal establishes is one Belgian buyer ^[2] and one product. It does not show predictive maintenance moving off failure-history models across the industry, and nothing in the announcement measures that. As Interesting Engineering reported, I-care said the deal expands its ability to serve sectors including defense and nuclear operations ^[6]. Founder and chief executive Fabrice Brion said the acquisition strengthens the company's ability to process difficult industrial datasets, and he pointed to the technology's potential in defense and nuclear environments ^[9]. Dogan Kaban, formerly Amiral's business development manager, said the combination creates an opportunity to take DiagFit to more industrial customers ^[12].

The setting that decides this sale is where the data gets processed. DiagFit supports on-premises deployment, and cloud deployment remains available for customers who can use it ^[10]. Defense programs and nuclear facilities can face strict security requirements around operational information ^[14]. For a site that cannot send operating data out, the on-premises option determines whether an evaluation starts at all.

Does the asset have a thick failure history, and is the operator permitted to move operating data off site? With a thick history and permission to export, existing supervised tooling covers the job. Thin history with no export is where a baseline-deviation product offers something a buyer cannot assemble from failure records they do not have, and that is where defense programs and nuclear plants sit ^[6][14]. A buyer also has to pin down which version arrives, since I-care said it will progressively integrate DiagFit into I-see, its AI-enhanced predictive maintenance platform ^[11].