Factories Are Teaching Old Machines to Talk

At an industry conference, Guidewheel CEO Lauren Dunford described how J&J achieved a 35 % productivity boost with AI and retrofit in a suture manufacturing operation after adding external sensors and data connections to machines that were already decades old. Dunford called the approach a “Fitbit for the factory,” a shorthand way of explaining how existing equipment can be monitored in real time without being replaced.

In that operation, the gains came from making machine activity visible and feeding that information into daily production workflows.

Johnson & Johnson’s example stands out for its reported scale, and as an example of making meaningful progress with AI in manufacturing; often beginning with collecting better data from the machines already in place. The process of digitally connecting older machines and using the resulting data as the foundation for analytics is central to what is often called Industry 4.0 or the Fourth Industrial Revolution, which emphasizes data exchange and smart technology in manufacturing.

Across much of the industry, the constraint exposed by cases like this is not a lack of equipment. It is a lack of usable information.

The Problem is the Missing Data

Industrial equipment is typically built to operate for decades, and many factories still rely on machines installed 20 or 30 years ago, particularly in pharmaceuticals, medical devices, automotive manufacturing, and food processing. Reviews of Industry 4.0 adoption note that legacy equipment often lacks the sensing and connectivity needed for digital transformation.

These machines often run reliably and meet quality standards, but they were designed long before modern analytics or cloud software became common.

As a result, many factories lack consistent, time-stamped records of what their machines are doing. That gap has become a practical barrier to using analytics and AI, which depend on clean, structured data to function. Peer-reviewed analyses of smart manufacturing adoption have identified data quality and inconsistent labeling as common obstacles to successful industrial AI projects.

McKinsey & Company has found that manufacturers that successfully apply digital technologies tend to focus first on improving visibility across shop floors and plant networks. In its analysis of Industry 4.0 programs, McKinsey reports that effective deployments are associated with large reductions in machine downtime, steady throughput gains, and improvements in labor productivity: outcomes tied to data and analytics more than equipment replacement.

Faced with this data gap, many manufacturers opt to retrofit existing machines rather than replace them outright. Replacing production equipment is capital-intensive and time-consuming, and in regulated environments it can also trigger lengthy validation and recertification processes. Retrofitting (adding sensors and data connections to machines already in use) offers a faster and less disruptive way to generate the information modern systems require.

Studies of so-called “brownfield” factories consistently identify retrofitting as the dominant path to digital upgrades, citing lower cost, reduced operational risk, and faster deployment compared with new equipment installations.

In practice, retrofitting usually involves attaching external sensors and edge devices that capture basic machine signals such as vibration, motor current, temperature, or whether a machine is running or stopped. Those signals are converted into standardized digital records that can be analyzed locally or sent to centralized systems. Because the machine’s control logic is left intact, the instrumentation can often be installed without halting production.

Making Machines Visible Changes How Factories Run

Once machines begin reporting events in real time, production reviews shift from reconstruction to diagnosis.

Downtime can be measured instead of inferred. Repeat stoppages can be traced to specific causes, and teams can prioritize fixes based on evidence rather than memory.

Studies of high-performing manufacturing sites support this pattern. Research on “lighthouse” manufacturers (production sites recognized for leadership in digital transformation) shows that real-time performance monitoring is often the foundation for improvement, linked to operational gains before advanced machine learning or autonomous systems are introduced.

Packaging and bottling operations in the food and beverage sector show how this visibility-first approach plays out in practice. These lines run repetitive processes where small interruptions can have outsized effects on output. Industry reporting suggests that instrumenting such lines often leads to faster root-cause analysis and incremental throughput gains, without changing the underlying equipment or introducing autonomous systems. Real-time monitoring and analytics are key components of the Industrial Internet of Things (IIoT), which connects sensors and machines for improved productivity and efficiency.

One common way manufacturers track these improvements is through overall equipment effectiveness, or OEE, a standard metric that combines availability, performance, and quality.

Making machine states visible allows those components to be measured directly, which in turn makes improvement efforts more focused.

Johnson & Johnson’s experience shows what happens when this data is integrated into established workflows. By connecting existing machines and feeding the resulting information into daily operations, the company was able to act on issues that had previously been hidden. The reported productivity gain reflects not just the technology, but how the data was used.

In manufacturing settings, this is what being “AI-ready” tends to mean in practice. It does not imply autonomous factories. It means having a reliable stream of consistent machine events that analytics tools can later use.

Most manufacturers don’t start with advanced models. They start by collecting enough consistent data to understand what their machines are actually doing. Predictive maintenance and quality analytics come later, once that history exists.

The early gains are reliable: clearer visibility, faster decisions, and fewer surprises on the factory floor. In plants with established daily reviews and continuous-improvement routines, instrumented data gets used immediately, compressing the time between a problem appearing and someone fixing it. Johnson & Johnson’s productivity improvement marks the high end of what’s possible.