AI gets real in managing asset performance

AI may be artificial, but it can deliver real improvements in multiple aspects of asset performance from reducing downtime to maximizing yields. Plus, once users recognize the advantages of artificial intelligence (AI) and add it to their activities, the potential paybacks are likely to be substantial.

“We’ve been working with enterprise AI for about 15 years, but we didn’t hear much talk about it elsewhere. They mostly talked about transitioning to the cloud and analytics, until November 2022, when every business suddenly wanted to know how to apply ChatGPT,” said Lila Fridley, VP and GM for reliability at C3.ai, during her presentation this week at the YNOW2024 users conference in Houston.

“Now,” she added, “it’s estimated that AI’s contribution to the global economy could reach $15.7 trillion by 2030 across industries such as industrial manufacturing, oil and gas, financial services, healthcare, defense and intelligence, government, utilities and retail.”

Use cases guide early steps

Because AI’s capabilities can be applied so widely, Fridley reported that users considering it should focus on specific use cases, which can guide them to the most helpful ways to apply it.

“We learned early that choosing the right use case matters a lot during the first steps towards using AI,” said Fridley. “A good use case can help AI provide more benefits, enable it to be applied across departments, and achieve higher-value results. Some good starting points include improving predictive maintenance, demand forecasting, reducing emissions and optimizing supply chain networks.”

After selecting a suitable use case, Fridley added it’s equally important to pick a reliable, scalable AI software platform that can add AI to use cases in a unified way. “Many users and suppliers try to stitch together AI models, but they all still run in isolation without a common work environment,” explained Fridley. “Ours are tools we spent 15 years building, that can work with others like Google’s large language models (LLM), integrate with existing systems and expose useful insights.”

Monitoring the monitors

Fridley reported that C3.ai’s software is a pre-built instantiation of multiple software suites, designed specifically for asset performance, supply chain, sustainability, state and local government, and defense and intelligence applications. More specifically, C3.ai Asset Performance Suite can reduce unplanned downtime by 50%, improve overall equipment effectiveness (OEE) by 5%, increase yields and total production by 2%, and reduce energy costs by 4%.

Users of C3.ai’s specialized applications include Shell’s industrial assets division, the U.S. Air Force, Cargill, and Koch Industries. “C3.ai is in 30 operating assets at Shell. We started at one site and scaled up from there to 10,000 different devices and processes that run 100,000 applications and AI models,” said Fridley.

“With the U.S. Air Force, C3.ai doesn’t just process data from sensors on aircraft. It also tracks parts and warehouse inventories to optimize its supply chain, and aid logistics by indicating whether needed parts can be flown to where they’re needed. This saves about $1.5 billion per year, and even increased the B1 bomber program’s availability by 25%.”