In the age of AI, data might be the most critical piece of the predictive maintenance puzzle. This could be a game-changer for manufacturers.
There’s no doubt predictive maintenance can unlock a lot of benefits on the plant floor. But the real key seems to be data.
In the age of artificial intelligence (AI), data collection is becoming easier, faster and more reliable. But many organizations are discovering just how little data they actually have to drive a fully functional predictive maintenance strategy.
Plant Engineering research shows that 67% of respondents in the 2026 State of Manufacturing Operations and Maintenance report said the emerging technology that is most critical to their facility’s success is predictive maintenance tools.
Another 49% reported that mobile maintenance tracking applications fit this bill, up 16% from 2025. Maintenance optimization driven by artificial intelligence (AI) was a key tool for 31% of respondents.
Data unlocks a strong maintenance strategy
When it comes to maintenance, the best strategies are driven by data, according to Limble’s chief executive officer Gary Specter.
“Data is the foundation of modern maintenance strategy,” Specter said in a recent Q&A with Plant Engineering. “Where schedules were once driven by manufacturer recommendations or reactive repairs, today’s most competitive manufacturers build plans around continuous streams of operational and production data. This ultimately transforms maintenance from a calendar exercise into a live, condition-based discipline.”
AI is reshaping how data is gathered
Data is the lifeblood of predictive maintenance, but AI is causing its own revolution in the maintenance space.
Ben Swisher, vice president and general manager of reliability solutions at Emerson, said modern industrial AI can parse data in a fraction of the time a human can.
“That has lowered the bar for entry into a successful predictive maintenance program –– if a team implements the right tools,” Swisher said in a recent roundtable for Plant Engineering. “Bolt-on solutions typically add more complexity, both in engineering connection to external systems and in training users to drive value from the systems. Fortunately, today’s most advanced automation suppliers are helping organizations navigate this challenge by engineering industrial AI directly into the tools reliability teams use every day.”
Edge environments, asset monitors and wireless vibration monitors, among others, use onboard AI and pattern recognition to deliver actionable information from raw data, Swisher said.
Generative and agentic AI offer even more assistance, said Scott Campbell, global leader of product management, environmental, social and governance asset management product success at IBM.
“GenAI chat capabilities using large language models are the most common starting point,” Campbell said in a recent roundtable discussion. “It is providing a new way to interact with asset and work order data, get quick answers to previously difficult maintenance questions and operation trends. Ironically, these AI interactions are also helping organizations realize, in many cases, they lack quality data. This inhibits their ability to go beyond inquiry into the higher-value insights and automation. It has shifted the focus for many to a data-first approach for leveraging AI innovations.”
Campbell said agentic AI workflows promise to unlock condition-based, predictive and financially optimized maintenance strategies that can deliver value to operations.
The benefits of better data
When it comes to reliability centered maintenance, manufacturers can find themselves at an advantage, if the proper data is collected, according to Javier Martinez, sales engineer with Teadit, in a recent Q&A for Plant Engineering.
“By combining field experience with empirical data, organizations are better positioned to optimize maintenance intervals, reduce unplanned downtime and improve overall operation efficiency,” Martinez said.
