Moet Hennessy is using artificial intelligence to detect wine defects before they develop, in a project with Analog Devices and the University of California, Davis.

The partners announced the effort on October 7. Their technology reads the chemical signatures given off by grapevines, juice and wine. Machine learning models then predict quality risks faster and more cheaply than standard lab tests.

The first success targets Fresh Mushroom Aroma, a flaw the wine world calls unacceptable in dry or sparkling wine. Researchers at Moet Hennessy’s research centre built a library of samples to train the algorithms to spot the warning signs.

The same platform could later detect vine diseases, assess soil and flag smoke damage from wildfires. The technology has roots in early research at MIT. The partners say the technology will keep human winemakers in charge, using AI as an early-warning tool rather than a replacement for their expertise.

Context: the champagne house, part of luxury group LVMH, has spent millions trying to beat the defect since the mid-2000s.