Turning Data into Gold: The 2025 AI Operations Playbook Five ways manufacturers are putting generative AI to work today By Derek Moeller It’s a story too many rubber processors are familiar with: A company makes a major investment in enterprise software, with the promise that digitizing a company’s maintenance, manufacturing execution, scheduling, and resource planning systems will yield major benefits in efficiency and profitability. But after thousands of hours invested in implementation, a general frustration often begins to rise in the company’s chief implementers when they realize something—staff aren’t using the system like they’d hoped. For instance, a problem happens with a production line, but there’s no work order for it. A work order is entered, but there’s little to no usable information in it. The system now has production data, but it’s not getting used to make decisions; instead, people are depending on what they heard from someone, or a hunch. The system offers a way to contribute process knowledge, but people don’t put it in. Information on procedures, even if they are in the system, don’t get used. The result is that companies have downtime because a machine doesn’t get fixed fast enough. Scrap is made because people don’t know how to fix a quality problem quickly. And trends in causes of problems don’t get used to address root causes of systemic issues. These issues happen in almost every implementation of a major enterprise software system. The user interface expert Jakob Nielsen, founder of the Nielsen Norman Group, tells us why. In his model, over the last century, software has had two “paradigms,” or models, of how the user has to interact with the computer. The first was batch processing. Think punch cards from the 1960s: A user creates a set of instructions by punching holes in cardstock, using hundreds or thousands of cards for a single program. They would then give this stack to the machine, which read the holes. There was no “using” the computer as we think of it today, no back-and-forth interaction, because a user had to wait their turn for the computer to run their batch. The second paradigm replaced batch processing, and is the one we’re still largely in today. It’s what Nielsen calls “command-based interaction.” A user tells the computer what to do via explicit command and gets a near-instant response. It was a huge advance from batch processing. AI & DATA STRATEGY 04 / INSIDE RUBBER / 2025 Issue 1
RkJQdWJsaXNoZXIy Nzc3ODM=