Leveraging AI to Leapfrog Industry 4.0

AI is a great equalizer, especially for manufacturing companies that fear they are behind the Industry 4.0 curve. In fact, if you believe your company has been lagging in that area, I’m going to suggest something to you that you might think is ludicrous; your company may be in a very advantageous position because it did not yet deploy (or attempt to deploy) an MES system.

Sounds counter-intuitive, I know, but I’ve observed it to be true in more than one manufacturing company in the last half year. Something very disruptive is indeed happening in the industry and it’s not a surprise that something is AI. But what is surprising is how it’s getting there. It’s not getting there through technology platforms; it’s getting there organically.

Technology experts and industry analysts have been touting AI as one more urgent reason to advance your company to an Industry 4.0 state, suggesting that if you don’t have a modern MES system, that you better hurry up and get one so you can catch up. The notion is that if you were lagging before, you are even more certainly going to be left behind now because “you’re not ready for AI”.

I found the opposite to be true, and here’s why.

Companies that for one reason or another didn’t deploy an end-to-end MES system in the last 5 or 10 years didn’t fade away and die, they continued running their business well and continued to meet the expectations of their customers, earning the continued business of their customers. I’ve often said, “the company with the best customers wins”, and the companies with the best internal processes and good people end up with the best products and the best customers. It’s never “the company with the best technology wins”. Technology is a commodity, and with AI, that’s even more true today.   

As far as plant floor technology is concerned, AI levels the playing field. It’s no longer who has the biggest engineering team or who has the biggest technology budget, it’s who has the best ideas, and who has the best access to their data.

The issue I have had with most established MES systems is that there is a premium set on knowing how to use the system. If you don’t know how to leverage that system, you become dependent on the technology provider, their channel partner, or one or two people in your organization that are proficient in their system. And that introduces a significant constraint in the continuous improvement and data driven decision loop.

AI democratizes the data analysis process and even software development.

People that know the manufacturing processes well need only a bit of AI aptitude, or they need to work in tandem with a clever young person that does. Manufacturers no longer need to rely on a single great programmer or have their own full stack development team. They need people with good ideas and where to find the data. There are little lakes everywhere in their operations.

About a year ago, I encountered a small niche manufacturer that gave up on trying to find an ERP and MES system that could replace their hodge-podge, ad hoc collection of software utilities they used to get their orders out the door. Most of their disparate systems collected and stored data, but there were no drivers or easy to utilize APIs to connect to the MES and ERP systems they were looking at. Their choices were to spend a lot of money having drivers built or start collecting data from scratch while learning and deploying an entirely new system.

Instead, they decided to use AI to build an umbrella layer that covered all their disparate systems, pulling the data from each system into it, without changing any of those systems. Over time, their plan is to incorporate the functionality of some of the stand-alone software utilities into the umbrella layer, eventually consolidating them all. They only have one person involved in this, and it’s all happening cost effectively and non-disruptively.

In the last 6 months, my company TPI-3 has worked with two other manufacturing companies (one large, and one small) that have chosen the same approach. Each of them was using accounting software initially developed in the 80’s or 90’s, and each of them had connected their critical machines and processes to stand alone or home built data collection systems. One of them had 7 different data lakes, containing years of relevant process and production data, but none of that data was easily accessible outside of the canned reports built into those systems. Through some experimentation with AI, they experienced a few very exciting early wins, which encouraged the management teams to give their people a broader AI mandate to see what they could do.   

Their people were very knowledgeable in their manufacturing processes and always had great ideas on how to correct or improve them. They didn’t have an MES because they were always too busy “getting the job done every day”, and as intriguing as the promises of MES systems were, the cost of the systems, and the time and effort required to deploy them, always kept them from going forward with them.

And this is where AI comes in. AI is rapidly making technology, and even software development, a commodity. Software technology never was one of the core pillars of successful manufacturing companies - even though many tech companies, industry experts and government incentive programs may have suggested it was. Good people, good products, and good customers - those are the pillars. Technology is only a tool, and now with AI, those tools are available to everyone.

Who is going to do best in the next half decade?

My bet will be on the companies with the best ideas and the best access to their data. You don’t need MES as a starting point; you need clever people that know your processes with some good ideas of their own, and one or two people with an aptitude and appetite for AI. And beyond that, they need the opportunity to see what they can do.

The more experts suggest it’s about technology, the more I say it’s about people. You don’t want to rent knowhow and insights from outside sources; you want to develop it and keep it in house. And it’s the same with your data; you don’t want to rent access to it, you need to own it, control it, and have easy access to it.

I think the “technology laggards” have always understood this.