Listen to the full interview!

Browse our ASSEMBLY Audible podcasts

The target audience is spot on for us in manufacturing. It's one we do no matter what our budgets are. It’s a high priority with some great returns. But what I find interesting is the subject of software. We were one of the few. You could count on one hand the software companies that were here, and they were scattered in subject matter. And today, there are nearly a dozen software companies here. And it's a dedicated category now. That's telling for today's subject and, overall, about how software has become a bigger piece of manufacturing, including in assembly.

A Brief History of Software in Manufacturing

Q: Let's talk about software in assembly. Can you give us a rundown of what that looks like and how it's changed over the last decade or so?

McKiernan: Software and manufacturing, if we look back a decade, there are a few major pieces, and now it's growing into some more subsets of each one of those major pieces. If you're a manufacturing company, you probably have back-office operations. And historically, that category is called ERP, enterprise resource planning.

That's your operations: ordering materials, CRM (customer relationship management), processing, billing, all day-to-day activities that go into running a business, and then moving them to the manufacturing floor, if that's what you do. Now, when you go out the back door to the manufacturing floor, you typically hand an order to the manufacturing floor. Without anything else, it would be processed manually through all the steps required to turn it into finished goods.

What developed over time was the next layer of software, called manufacturing execution software. There's ERP and MES, and for the majority of the last decade, that was it. There was some continuing evolution of each of those layers, improvements, and granularity through those, like subject matter focus specific to different types of industries and manufacturing. That's been eight of the last 10 years. 

The Connected Worker Adds a Layer to the Stack

More recently, the connected worker has come into play. Connected worker is a way to touch every team member in the organization, from back office to shipping and receiving, to maintenance to the operator and station, and even between when they're working, from when they're going from their station, maybe to a break, or they run into a challenge on the floor.

Connected worker is a method to empower and engage every employee in the organization with bidirectional communication and information at their fingertips in an easy-to-use, granular way. So now we have ERP, MES, and this new frontier of the connected worker. If we drew a Venn diagram, all of them would overlap to some degree.

There's some discussion around each one of those. But for the most part, those are the three (or four now, including AI) layers in play. The overriding subject that you're going to hear about is artificial intelligence, which is affecting every single layer from top to bottom.

How the ERP Layer is Benefitting from Artificial Intelligence

Artificial intelligence will continue to evolve. Where we see the easiest and quickest returns on our efforts are in back-office operations, the ERP layer. Because so much information is concentrated there. There are so many workflows that can be aided by artificial intelligence agents specific to these neural subjects, which they're good at right now.

What we're seeing today is that, in our company, the ERP layer is benefiting from artificial intelligence. We are creating narrower, more specific agents that work within process flows for orders, such as sending RFQs and processing. And that's a very arduous process. If you have to send RFQs, write them and send them to your supplier base.

It's a very arduous process to review them and ensure all the information you asked for is provided, but that's easy for artificial intelligence to take over. Now you're just supervising it and setting the stage for it, just as an example. But think about all the layers that happened in the back office. They're really good targets for artificial intelligence. 

AI on the Factory Floor

Now we start moving out onto a manufacturing floor. It gets a little more challenging to find subjects where we can affect the bottom line. But there are certainly many ways artificial intelligence can help in the manufacturing execution layer, such as reporting, visualization, and reviewing data to identify constraints in quality or throughput.

Anything that you can process quickly and easily that you used to have to do through mining, looking at it, and reviewing it. Re-processing that data to see if you can find attributes that are looking for a trend that things are going out of tolerance, for example, and a quality point of view, or you're seeing a lot of variability in process and throughput. Your station cycle times vary quite a bit. These are easy things that the data set being created for manufacturing, and the manufacturing execution layer can look at to provide you with insight quickly. There's a category in artificial intelligence called knowledge management. It basically takes all the how-to manuals and what-if information you would normally have to find or search for, and, just like the chatbots we're working with, ChatGPT, it can search the internet through natural language and by asking questions. For example, narrow the subject down to all your core information for your products on the floor, and it'll work through that environment to give you this answer in a much faster way than you would before, where you have to sift through stuff for hours or just the fact that you have to open up a document, go through the appendix, and find the subject you're looking for through natural conversation. You can tell a chatbot what you're asking for, and it returns the answer to you easily.

Q: What you just described is essentially the new manufacturing stack?

McKiernan: Yes, that's what we refer to as the new stack. Stack is a commonly used term in the back end of software development: the components that make the software flow are called the stack. 

Q: Why does manufacturing today need a stack rather than a patchwork of tools? 

McKiernan: Ideally, what we used to have was vendors for each of the categories we've been talking about. There are several vendors in the ERP layer, several in the MES layer, connected worker, and so on. When you buy and source these software categories independently, you basically have to learn how to use each one. You probably need experts who are specialists in implementing it and continuing to maintain it for each one of these.

In theory, the stack becomes the products that are required. But if we could get something that was more integrated between all of them, instead of having five vendors in each of these categories. One vendor is building, buying, and integrating all three layers into one solution for customers in manufacturing.

So that's where it starts to be advantageous: you don't have to source an ERP and go through all the cycles and effort required to find a vendor that aligns with your needs in your unique, specific manufacturing segment. Same thing for MES; same thing for connected work. What I would recommend is to look for vendors who are thinking, Is that a complete deliverable for you?

A stack that's integrated seamlessly, beginning to end. So you have one vendor to work with that's specialized in your vertical manufacturing, from the back office all the way through pack out.

Q: Why does integration matter more for mid-sized companies than large enterprises?

McKiernan: Some of these are pretty obvious. Large enterprises have resources beyond those of even large companies. So, enterprise, large, mid, and small. When you work with large enterprises, they typically have the IT infrastructure, software infrastructure, and resources dedicated to making this. But just because of the sheer size and scale of their work. When you move down to even large and mid, it's important to make the right choice because the expense, they just flat out don't have these people on staff, they don't have the infrastructure to support what's required to implement, deploy and maintain these types of applications.

That should go into the decision-making when you're considering what vendor to choose. You could have seamless integration, which really reduces the workload and the need for the different disciplines that are typically available in a large enterprise. 

Q: What are some of the risks or challenges of bolting together multiple solutions instead of using a unified stack?

McKiernan: You have a process to go through and a vendor that has a solution for you. That cycle involves a lot of effort. 

Ideally, you look for a seamless, integrated solution that lets you work with a single vendor with the same ideas, look, and feel. All the connectors between these layers are implemented in a highly integrated manner, rather than having customized patchwork between each layer.

Q: How does integration between ERP, MES, and the connected worker tools change daily or improve daily operations?

McKiernan: With the advances in artificial intelligence and the ability to have insights into all the data produced by these software layers, you have faster decision-making, fewer errors, and better visibility throughout your operations. Overall, more efficiency. You're more competitive as an organization and more likely to win business that you probably weren't in line for before, thanks to these types of insights.

Q: Is this Epicor's current focus?

McKiernan: Yes, it's very near and dear to us. We have a product in the ERP layer. That's our flagship product, Kinetic. Kinetic is where we're starting with our largest emphasis in artificial intelligence. We have won some awards, but we're now working in the MES layer, connected process control.

And we have a new layer of connected worker. All three of those are being worked on, connected, and integrated seamlessly. It's a big advantage to the customer. It should be something considered for the industry as a whole.

Q: How can companies ensure their stack grows with them as they scale?

McKiernan: That comes down to selection. You want a company that can take you from where you are today and scale to where we hope to be in five to 10 years. And that's important in general; some companies specialize in the top end of the ERP and MES market. 

And they typically can't come down to the mid category at all. Now we have the space at Epicor to be well positioned because we can go down to small without any problem, and we have customers that are in the enterprise category as well.

It's important to dig in, ask tough questions, and request customer references in your space. Ask for references for where you hope to be in the next five to 10 years, so these products and companies can scale with you.