AI & Automation: Real Tips From Real CEOs

American factories and warehouses are undergoing an automation boom that was well on its way before the pandemic, was supercharged by the virus, spurred further by supply-chain debacles and generational labor shortages, and now is entering a new phase in which companies’ need to reshore industrial operations has become another driving force.

Sure, the arrival of ChatGPT has created existential questions for everyone from lawyers to artists to financial planners. But generative artificial intelligence is already insinuating its way onto the factory floor as well. Other hot areas of manufacturing and distribution automation include the creation of digital threads and “twins” for simulating and improving processes, and human-machine collaboration.

“This is a moment for real acceleration of automation,” says Asutosh Padhi, managing partner in North America for McKinsey. “There’s a felt need today, driven by talent shortages and investments in digital, that a lot of companies have done.”

Slim worker availability and high labor costs continue to dominate CEOs’ reasons for accelerating automation. A recent Chief Executive survey found that 51 percent of companies with more than $250 million in revenues are investing in automation to reduce the number of employees in the back office in response to rising wages, and 44 percent of those companies are doing the same in the plant.

AI is, of course, the leading edge of this phenomenon, both in generative and preexisting forms. It’s entering factories and warehouses rapidly, especially in areas of input optimization, maintenance and relationships between product design and manufacturing execution, including greater customization.

“The industrial use of AI has drastically shifted how manufacturers approach, think about and execute business processes,” says Rachael Conrad, vice president and general manager of customer support and maintenance for Rockwell Automation. “Through the simplification and automation of data analysis, manufacturers can easily raise awareness to areas that need improvement.”

Vijay Sankaran, chief technology officer for Johnson Controls, says industry is “well beyond exploration of AI as a sort of useful capability. We’re at the point where AI is becoming embedded in solving problems for almost everything that we do. It’s quickly moving from experimentation to much more broadscale adoption.”

Machine learning, video analysis and other tools required for AI have been coming together for a while. The last pieces to fall in place are “the cloud and 5G,” says Suketo Gandhi, digital supply-chain leader at Kearney consultants. “You can suddenly raise the efficiency of individual workers and factories.”

Keeping machines going through process analysis, troubleshooting and predictive maintenance is one huge application area for AI, enabling factories to far exceed the capabilities and pace of humans alone. “Often, AI is able to use sensors to identify something the humans are hearing or seeing or smelling for when that person adds the catalyst to a process,” says Alice Globus, CFO of industrial systems supplier Nanotronics, which has been providing AI to manufacturing clients for more than five years. “Also, if a valve is locked open, it could take a human process engineer hours to figure out where it is, whereas AI could immediately identify which valves and sub-processes need to be looked at.”

 Also, AI-powered “recommendation engines” can plow through contingencies such as material shortages and machine faults. “The person getting the most daily value from this sort of AI is not a data scientist or even a manufacturing engineer, but the operator who is fine-tuning and making adjustments to a machine,” says Stephen Laaper, smart manufacturing leader for Deloitte consultants. “It helps them make the next best decision in the execution of their job: ‘Please choose A, B or C.’”

Predictive maintenance is another fast-growing area for industrial automation. “AI is starting to be built in, to help technicians monitor anomalies,” says Blake Griffin, senior analyst at Interact Analysis. “For a long time, you had to rely on subject-matter [human] expertise they developed over years and years. But now AI is a huge play in keeping equipment up and running and healthy, reducing unplanned downtime.”

Human-machine collaboration in general is a strong theme running through new automation efforts, including “collaborative” robots that work alongside and in some cases even around human workers. Collaborative robots are “easier to set up, cost less, are safe and don’t require as much floor space or the same level of engineering expertise” as fully autonomous robots, notes Jeff Burnstein, president of the Association for Advancing Automation. They also offer a path for small and medium-size companies to ease into robotics applications.

For instance, in the early days of robot installation in automotive plants, it was risky to life and limb to assume the machines could work in physical concert with people. But in one of today’s newest assembly plants, the factory in Dearborn, Michigan, where Ford builds the new F-150 Lightning all-electric pickup truck, human-robot collaboration, in an intricate ballet weaving around the truck—is a staple of the process and a huge efficiency boost.

Yet in a different way, humans also remain a big impediment for the spread of AI. “More sophisticated automation requires more technical resources,” Laaper says. “But we’re tapping into an increasingly scarce skill set of robotic engineers and technicians.”

Costs of implementation remain another formidable obstacle to continued momentum for industrial automation, including initial affordability and generating an acceptable return on investment.

This is a problem especially for smaller manufacturers. The Chief Executive survey found that only 28 percent of companies with under $10 million in revenues are investing in tech and automation on the front lines, compared with 50 percent of companies with $1 billion or more in revenues.

And in a survey by Visual Components, 43 percent of manufacturing decision-makers said between 21 percent and 60 percent of their operations are being held back from being updated with new technologies due to continued cost pressures.

Still, Christian Hasenoehrl, a global leader for consumer and industrial accounts at the Korn Ferry executive-search firm, says CEOs also are haunted by one thought. “If your output is only at 95 percent or 90 percent of what you’re capable of,” he says, “you can be leaving billions of dollars on the table if you don’t automate.”

Maker of modular buildings

DESIGNING BUILDINGS TO FIT

Suchit Punnose, CEO
Modulex • Pune, India

We use AI, IoT and blockchain to design and site modular buildings that we construct and erect all over the world. They include the tallest modular hotel in New York, a Marriott. We build modules in the factory and then ship them, right down to the bed, mattress and pillows.

So you can’t automate processes inside a room with plaster walls. How we do apply AI is in using our design library of different types of buildings to design structures to their sites in just a few seconds, including reflecting local planning regulations. We also use AI to improve the design efficiency inside our rooms. And we use AI in our factory to optimize supply chains, predict demand, improve our delivery times and help with costs.

Diversified contract manufacturer

TRIMMING COSTS THROUGH AI CALLS

Paul Baldassari, Executive Vice President of Worldwide
Operations, Flex •  Singapore

In our hundreds of plants, we are building many use cases for AI. For instance, in our large factory in Guadalajara, Mexico, 10 percent of indicated failures weren’t actually failures. People had to check each “failure” and manually override it. But with an AI application, we only have a .02 percent false-failure rate. And now one person can actually monitor multiple lines and focus on the hard cases. A job that was boring and repetitive has become a higher-skilled job that requires more flexibility and a more innovative human mindset.

Also, in the past, we ran preventive maintenance on laser cutters and pulled out a unit every couple of hours. Now, with an AI solution, we can just see how much electricity has been used by that unit and whether we need to do maintenance or not. That also detects errors we may miss during regularly scheduled maintenance. So we save on spare parts and increase line reliability, up time and product quality.

Maker of food-processing equipment

KEEPING HUMANS IN THE LOOP

Daniel Voit, CEO
Blentech • Santa Rosa, California

On the production floor, you want to have human judgment at critical points. You need to be able to parse where a human needs to make a decision and break up those activities into automation, decision, automation. You can automate one or two processes without automating the whole line.

For example, with our equipment, some customers will want to make a mirepoix or a French onion soup or sauce, and those processes are inherently difficult to automate. As you change from one product to another, you need a configurable framework for introducing automation as well as human steps. It may require human oversight to a certain point and, after that, to take over with automation.

Contract electronics manufacturer

SEEKING NEXT-LEVEL AUTOMATION

Misha Govshteyn, CEO
MacroFab • Houston

We’re in an industry that has been highly automated for a long time, so we’re dealing with constraints in advancing it. One area that we’re still trying to automate is secondary assembly, where you take a circuit board and put it in an enclosure of some sort. The problem is that enclosures and mechanical parts come in all shapes and sizes, so it’s a lot more expensive to automate. It’s not making major inroads.

Also, we have robotic automation, but what they’re doing is still fairly big and doesn’t require fine-grain motor functions. The fingers and motor functions just aren’t small or good enough to weave wire harnesses together, for instance. When robotics get nimble enough, at some point you may be able to use machine learning and AI to teach them to go build something.

Third-party logistics provider

AUTOMATING THE WAREHOUSE FLOOR

Kristi Montgomery, Vice President of Innovation, Research and Development
Kenco • Chattanooga, Tennessee

Distribution is starting to catch up to where manufacturing has been in automation. Amazon was among leaders with autonomous mobile robots in its warehouses several years ago, and after Covid hit, we saw an enormous uptick in automation to address labor shortages and space constraints.

In our 100-plus facilities in North America, we also have more semi-automated augmentation of human labor, such as mechanical container-unload devices that help people palletize boxes and solutions that create the right box size on demand, reducing the sizes of small-parcel shipments. With UPS and others looking at dimensions as well as weight, it becomes really important to optimize that for customers.

Recreational vehicle maker

LEVERAGING PRECISION ON THE LINE 

Huw Bower, President
Winnebago Brand of Winnebago Industries • Forest City, Iowa

We’ve invested around specific forming and cutting machines that cut with speed and volume but also accuracy, and that makes the whole assembly process much easier. On the line and in assembling cabinets, it’s a craftsman environment, where they’re highly skilled, and it’s difficult to automate. But because this automation can control tolerances to a much greater degree, the fit and finish becomes infinitely easier, and the quality is higher.

We use AI increasingly in digital work instructions and for the efficacy of training. We are also digitally transforming our operating environment and implementing a new ERP system in our core motorhome business in Iowa, so our advanced technology group and center of excellence are looking at AI opportunities in the volume work in our plastics-thermoforming area and repeatability in stitchcraft—integrated vertical operations where we could leverage insights from AI.

Climate control systems provider

KEEPING AN AI EYE ON FACILITIES

Vijay Sankaran,  Chief Technology Officer
Johnson Controls • Glendale, Wisconsin

We’re at a point where AI is becoming embedded in solving problems for almost everything we do. For example, we use generative AI for a training set around our internal technical documentation that is pretty game-changing in terms of technicians asking how to fix things. It’s sort of like YouTube [instructional] videos.

Under our OpenBlue system with our clients, we’re connected and constantly seeing data and using AI algorithms to monitor signals and faults from chillers, and predicting when they might have an issue so we can safely shut them down. We use deep neural networks and image recognition, for example, to identify flares from client smokestacks to indicate particular warning signs about the health of that facility.

In our manufacturing, we use AI to help meet our net zero emissions targets, starting with eight of our facilities to baseline our carbon emissions and leverage AI to make adjustments in real time based on things happening in that facility or different external environmental conditions.

Climate control systems provider

KEEPING AN AI EYE ON FACILITIES

Vijay Sankaran,  Chief Technology Officer
Johnson Controls • Glendale, Wisconsin

We’re at a point where AI is becoming embedded in solving problems for almost everything we do. For example, we use generative AI for a training set around our internal technical documentation that is pretty game-changing in terms of technicians asking how to fix things. It’s sort of like YouTube [instructional] videos.

Under our OpenBlue system with our clients, we’re connected and constantly seeing data and using AI algorithms to monitor signals and faults from chillers, and predicting when they might have an issue so we can safely shut them down. We use deep neural networks and image recognition, for example, to identify flares from client smokestacks to indicate particular warning signs about the health of that facility.

In our manufacturing, we use AI to help meet our net zero emissions targets, starting with eight of our facilities to baseline our carbon emissions and leverage AI to make adjustments in real time based on things happening in that facility or different external environmental conditions.