Your Utility Has the Data. Is It Ready for AI?
Part 4 of our Driving Digital in Water and Wastewater Operations series
AI has quickly become part of almost every conversation around the future of manufacturing and utilities. There is no shortage of ideas about what it could do, from predicting equipment failures and reducing energy use to improving quality, forecasting demand, and helping people make faster decisions. But for organizations that have spent years investing in SCADA systems, historians, sensors, reporting, and other operational technology, there is a more practical question worth asking first: Is the data you already have ready to support what you want AI to do?
That question came up during our webinar, Driving Digital in Water and Wastewater Operations, when an attendee asked what else could be done with years of historian data that was primarily used for trend analysis. Scott Lamothe’s answer was simple: having that historical data is the first step toward analytics and AI. Without it, AI lacks the operational history it needs to understand what has happened, identify patterns, and answer a specific question.
It was a good reminder that getting value from AI doesn’t necessarily start with an AI project. For many manufacturers and utilities, it starts with looking at the information they’ve already been collecting differently.
Start with What You’re Trying to Fix
This may be one of the most important points Scott and Kurt make during the webinar. Once the data is available, the next question isn’t which AI tool to buy. It’s what problem you actually want to solve.
Scott suggests looking at the top business objectives already on the table. It could be reducing energy consumption or chemical usage. It could be less downtime. In manufacturing, that same thinking could apply to reducing scrap, improving throughput, understanding process variability, or finding the conditions that consistently produce the best quality. The problems aren’t new just because AI is.
That’s what makes this approach practical. Instead of creating an AI initiative and then searching for a use case, start with a problem the operation already needs to solve. If reducing energy costs is a priority, what can years of operating history tell you about when and where energy is being consumed? If downtime is the problem, what patterns show up before equipment or processes begin to struggle? If quality is inconsistent, what was different during the periods when the process performed at its best?
Those are questions operations teams have been asking for years. AI and analytics give them new ways to find answers.
The Data You Already Have May Hold the Answer
The webinar gets particularly interesting when Scott talks about using historical data to identify periods when an operation achieved a desired outcome. His water example examines whether there were times when water-quality requirements were met with fewer treatment chemicals. Instead of asking someone to sift through months or years of information manually, analytics can sift through that history to identify when it happened, what conditions were present, and whether those conditions could be recreated.
Kurt expands that idea beyond chemical usage. He talks about using operational data for leak detection, energy consumption, and demand forecasting. If a utility can better understand how much capacity it will need tomorrow or next week, it can make different decisions about when to run energy-intensive equipment.
Manufacturing teams face very similar questions. Think about a production line that has been running for years. Somewhere in that history are your best runs, your worst runs, quality deviations, downtime events, energy spikes, maintenance activities, and changes in raw materials or operating conditions. The value isn’t simply having that history. It’s about being able to put it into context and use it to understand why one outcome differed from another.
Being Data Rich Isn’t the Same as Being Ready
Earlier in the webinar, Kurt shares something he heard from a customer: the organization was “data rich” but “action poor.” They had dashboards, alarms, reports, and plenty of information, but there was still a disconnect between having the information and knowing what to do with it.
That distinction becomes even more important when AI enters the picture. Years of historical data are valuable, but simply having a large volume of data doesn’t automatically make it useful. Operations still need access to the right information, enough history and context to understand it, and a clear objective for what they want to improve.
There is also the question of where that data needs to go. Later in the webinar, Scott and Kurt discuss the relationship between on-premise operational data and enterprise data architectures. Operational systems may need to remain protected within the OT environment, while selected data also needs to be made available to cloud platforms, enterprise analytics, modeling tools, or AI. That balance between access and protection is another piece of building a usable industrial data foundation.
AI Should Still Lead to an Operational Outcome
With all the attention AI is getting, it can be easy to make the technology itself the goal. The webinar takes a much more practical view.
Scott puts it in terms almost any plant or utility can relate to: identify two or three business objectives first. Energy savings. Chemical savings. Less downtime. Then determine whether analytics or AI can help you get there. He makes the point that this applies whether you’re running a manufacturing facility or a municipality.
That’s also a useful way to measure whether an AI project is worth pursuing. If the outcome is lower energy use, improved reliability, less waste, better quality, reduced downtime, or more informed decisions, you have something tangible to work toward. If the only outcome is being able to say you’re using AI, it may be worth going back to the problem you’re trying to solve.
Your Data Foundation Comes First
The good news is that many utilities and manufacturers aren’t starting from zero. Years of investments in historians, SCADA, automation, sensors, and reporting have created a tremendous amount of operational history. The opportunity now is to make that information more usable and connect it to the problems that matter most.
That doesn’t mean every organization is ready to jump directly into AI, nor does it mean AI is the answer to every problem. It means there may be more value sitting inside your existing operational data than you’re using today.
This is where having the right data strategy and operational foundation matters. At InSource, we help industrial organizations examine the systems and data they already have, identify information gaps, and determine practical ways to turn operational data into usable insights to improve performance.
Watch the On-Demand Webinar
Throughout the webinar, Kurt Peterson and Scott Lamothe discuss what they’re seeing across the utility industry, including cybersecurity and disaster recovery, workforce and resource challenges, operational data, analytics, and AI.
If you’re sitting on years of operational data and wondering what more you could be doing with it, watch the on-demand webinar to hear the full discussion and the real-world examples behind it.