Wine Industry Partners Unveil AI System That Detects Defects Before Standard Tests

Moët Hennessy, Analog Devices and the University of California, Davis said Monday they are working together on an AI-based sensing system designed to detect chemical signals from grapevines, juice and wine earlier than standard testing methods, with the aim of spotting quality risks before they become obvious in the cellar or the bottle.

The partners said the first result from the effort is an early warning system for a defect known as Fresh Mushroom Aroma, or FMA, a problem that can alter a wine’s smell and reduce its market value. According to the companies and the university, researchers were able to identify samples that had a higher risk of developing that defect before it could be detected through conventional methods.

The project combines UC Davis expertise in viticulture and enology, Moët Hennessy’s wine research and production data, and sensor and machine learning technology from Analog Devices, known as ADI. Moët Hennessy is the wine and spirits division of LVMH. ADI said the platform analyzes a broad range of chemical information and uses AI to determine which signals matter most, rather than testing only for a fixed list of known markers.

The work builds on earlier research that began at the Massachusetts Institute of Technology, according to the announcement. ADI said it has continued to develop the technology for chemical signature analysis across different fields and is now adapting it for vineyards and winemaking with the other partners.

At the center of the first demonstration is a sample library created by researchers at Moët Hennessy’s Robert-Jean de Vogüé Research Center. That library, along with associated production and quality data, was used to train machine learning models to recognize chemical patterns linked to FMA risk. The organizations said that gave the system the ability to flag samples at an earlier stage than usual, which could allow producers to act before the defect becomes established.

That matters because quality defects in wine can be costly. If problems are found late, producers may have fewer options to correct them, and some batches may lose value or need to be downgraded. Earlier detection could help wineries reduce losses, improve quality control and make decisions sooner about fermentation, blending, storage or other interventions. For the broader beverage sector, the same logic could apply anywhere producers need to catch subtle quality changes before they affect a finished product.

The partners described the technology as a decision-support tool, not a replacement for growers or winemakers. They said the goal is to give human experts more information at the right time while leaving final judgment to people in the vineyard and winery. That point is important in a field where production choices often depend on experience, local conditions and sensory evaluation, not only lab data.

The collaboration also points to a wider use of AI in agriculture and food production, where companies are looking for faster and less expensive ways to monitor biological systems. In this case, the system is focused on volatile compounds and other chemical signatures emitted by living material. The idea is that those signals can reveal biological changes or stress before visible symptoms or traditional quality problems appear.

Beyond FMA, the three organizations said they are studying other uses, including early detection of vine diseases, soil assessment and identification of defects linked to climate-related events such as wildfires. Those applications remain in development, but they suggest a larger role for the platform in vineyard management as climate variability increases pressure on grape quality and consistency.

Potential wildfire-related uses could draw close attention in California and other wine regions that have dealt with smoke exposure and crop losses in recent years. A system that can detect chemical changes tied to environmental stress at an early stage could become valuable not only for premium wine producers but also for growers trying to manage risk across larger volumes.

Ben Montpetit of UC Davis said in the announcement that the resilience of the grape and wine sectors will depend on stronger knowledge and people who know how to use it. He said the collaboration is intended to advance both, by building tools that improve decision-making and by helping train the next generation of viticulture and winemaking specialists to interpret the new data.

Manuel Reman, a member of the Moët Hennessy Executive Committee who oversees research and development, said the effort could help producers detect and interpret the “chemical language” of grapes and wine before quality problems are visible. Max Shulaker, chief of ADI’s Health Solutions business unit, said the project joins AI-enabled sensing with viticulture to read signals emitted by living systems and shift quality management from reacting to anticipating.

ADI said it ultimately plans to make the technology available for wider industrial use. At the same time, Moët Hennessy and UC Davis said they are exploring whether shared reference libraries could be developed to benefit the broader grape and wine industry. That could matter if the system proves useful across different regions, grape varieties and production methods, since AI tools tend to improve when they are trained on larger and more diverse data sets.

No timetable was announced for commercial rollout or wider industry access. The organizations also did not disclose financial terms of the collaboration. What they did present was an early example of how semiconductor-based sensing, machine learning and agricultural science are being combined in a sector that has traditionally relied on slower lab analysis and human sensory assessment.

For wine producers, the immediate promise is practical: earlier notice of a defect that may otherwise appear too late to manage efficiently. For the beverage industry more broadly, the project is another sign that AI is moving beyond marketing and logistics into the chemistry of production itself, where even small gains in early detection can have an outsized effect on quality, waste and cost.