Why Your AI Gains Disappear

Ed Ballina (00:00)
Hi, I’m Ed Ballina.

Alvaro Cuba (00:03)
Hello guys, Alvaro Cuba here.

Ed Ballina (00:06)
And welcome once again to our Manufacturing Meet Up Podcast. This is a show where we kick back on our downtime, we get real about efficiency on the plant floor and exchange our experiences. So, join us.

Alvaro Cuba (00:26)
So guys, for today’s episode, ⁓ with Ed we were thinking about which topic to bring. And we have been having different questions from you on the same topic. And also when we go to the plants, is this question, okay,  AI and automation is coming and it’s clear the benefits and the impact.

But how we translate that into business gains, concrete business gains. And for sure you have heard this term of productivity leakage, which is when you get the benefit, but somehow it doesn’t translate into a business gain. So  we check a couple of statistics. This one, is from Gartner 2025, the CEO and the Senior Business Executive Audit. It’s specific for more for office or administrative, but I think it helps to exemplify it. It says that AI is saving 5.7 hours per employee per week in an office environment. And 1.7 of that, of the 5.7, 1.7 is going into high value business gains. 0.8 goes into solving things that AI couldn’t solve. And the rest is a big question mark. So that, it’s happening there and it’s easy to quantify.

But with Ed, that resonated to us. We were saying, yeah, we are seeing exactly the same, this productivity leakage in the plants. And when Microsoft did another study, 2025, the CEO study, and says that CEOs, 34% want to boost productivity. But 43% want improving decision-making.

And 71 % wants strong innovation and create new products and growth. So as you can see, there is different business priorities depending on each of the companies. So we said, OK, let’s take a little bit of time and use a little bit of our experience of how we can  identify these leakages.

Or on the positive, the best practice on how you can use artificial intelligence or automation gains fully into the business goals. So, Ed

Ed Ballina (03:36)
Yeah, there’s quite a bit to unpack there.  But I think one of the issues that we’ve been struggling with  is AI as it came upon the landscape, right, has been a huge splash, right? And you can say AI, can say IoT, you know, industry 4.0, they, the terms are used somewhat interchangeably. They’re all kind of talking about the same thing. And to be honest, a number of companies adopted it because it sounded sexy. If you weren’t doing AI, if you didn’t tell your shareholders, you had some IoT platform work going on, it felt like you were behind the times. Sometimes companies went into these initiatives without a clear understanding of how it was going to all flush to the bottom line in terms of outcome. They focused on actions and activities that did not always, to Alvaro’s point, yield a tangible  result. And let’s not kid ourselves, we’re doing this because there’s tremendous opportunity, right, for productivity, for both personal and business enhancement. But at the end of the day, this has to pay for itself and then some. And as Alvaro and I have talked about this,  we’ve kind of broken down the leakage into two buckets. One, process machine. And one for me? ⁓

Alvaro Cuba (04:57)
One for you, one for me.

Ed Ballina (05:01)
We’re always about sharing, you know, we share buckets. And we’ve also realized that we’re better when we share our individual buckets with each other than when we just talk for four minutes and then go, and now it’s your turn. So I’m to kick it off by talking about process and machinery,  impact or leakage and then Alvaro and I will also talk about people, right? So you’re going to get a little bit of exchange. And so.

Alvaro Cuba (05:02)
Yeah.

Ed Ballina (05:29)
Let me give a couple of examples of where this product, the, the productivity leaks out when it comes to machineries. And that is I’ve seen companies that invest a lot in AI in one particular piece of equipment and that particular piece of equipment, may be the heartbeat of their process, right? We’ve all learned about the theory of constraints, right? You really got to worry about that machine. That is the constraint for the whole line. And we typically focus on that for good reason. However, if you stop there and you don’t improve the reliability of your whole line, you are not going to get the full benefit. So case in point, I know a little bit about bottling. You can go in and spend a lot of time improving the performance of a filler, reliability, speed, quality, right? And maybe you get five, 10 % speed increase and you feel really great. But as you start pushing that down the line, all of a sudden you find out your downstream equipment can’t keep up with it.

And you’re starving the filler because the upstream equipment can’t either. And at the end of the day, you know what you find at the end of your palletizers? A big fat zero sometimes. Most of the time the leakage happens so you pick up two or three percentage points where the potential was five or 10. So that’s an example where if you don’t focus on your system, which is your whole line, you can improve one single machine, but it’s not going to translate the way you would like it to.

Alvaro Cuba (06:57)
And allow me to build on that. Even when you get finally the entire machine and you start gaining output, if you don’t see the whole picture to Ed’s point, in this case, if you don’t bring the planning area into that, you could get the benefit of additional speed and even additional product on product that the company doesn’t need. Or there is too much stock or there is not enough sales. So that’s another big area of leakage. As you get your lines to perform better, it’s planning the ones that need to factor that into what to do in a way to benefit the business the most.

Ed Ballina (07:56)
Alvaro, that is so on point. Even before we talk about AI, back in the days where I was involved doing line efficiency work, and you could literally take a line from 60 % efficiency to 80 % efficiency over a course of nine months if you have a good approach and you’ve got access to capital and a little bit of expense. Anyway, one of the biggest sins is if you increase output of a line, even if it’s 10%, right?

And you don’t figure out what to do with that product. Here’s how you reward the operators and mechanics of that line for their work. You send them home early because there’s no product for them to make. How soon do you think they will figure out that all those gains are costing them dollars in their pocketbook, right? And rightfully so. They will start slowing things down to meet the status quo again. So to Alvaro’s point, if you don’t let planning right? Supply chain planning know that I have this excess capacity, let’s put this to good use. All of a sudden you fill out your warehouse with the product and they tell you, hey, you’ve got to shut down for a week, my friend, because you’re making too much product. That is the death knell.

Alvaro Cuba (09:08)
And that also generates on the other side, leakage because nowadays one of the biggest benefits of AI is in planning. Remember the planning, how difficult, how many SKUs, how many hours, how many  things moving up and down, it was impossible to do a re-planning.

One per week and it was a lot. Once a month for the entire business, it was a lot. With AI right now, you can re-plan any time, five, 10 times a day if you want it, just to maximize it. But if you don’t have this connection between what the planning can do and the operation can do, then you end up having leakages in both sides.

Ed Ballina (10:04)
Absolutely.  So now let’s talk about one example of something that does work. Okay. This is not really new, new technology, but  in many of our, if you were direct store delivery,  you pick a lot of cases, you hand pick a lot of cases, especially with a lot of SKUs right? So the way this works is the man comes in, the warehouse creates a pick sheet for people to actually build pallets of these products.

And because everybody wants to have their own individual flavor and all that, we have a thousand SKUs. So we have folks that are touching 2000 boxes a day, right? To build a pallet that show up in the marketplace. And it was all done with a piece of paper and some checkoffs. Eventually figured out that if we could do voice technology, to lessen the, you know so they would wear headphones, you’d pick up  the…demand from the system, they would tell you go to site two, pick four cases of that, right? And as you did that, as soon as you did, you confirmed, pick two cases, SKU ba ba ba ba. It really reduced a lot of errors. It made the pickers a lot more efficient  and was definitely an outgrowth of technology. Now you apply AI to that and you start doing dynamic routing of loaders through a pick area  or in the old days we used to route every night, right? The trucks would go out, right? With a fixed route. Now AI can optimize your delivery system to get you better stops and better quality and interaction with the customer. So a couple of examples there.

Alvaro Cuba (11:48)
Yeah, yeah, very valuable example. another one is the KPIs, you know, and how you manage your KPIs in a way that you get really the benefits into the business. Ed was, was talking about fillers, bottle fillers. I can, I have the same experience in, in chocolate or cookies or crackers, packaging lines and  producing lines. If you focus on meantime between failure or next time to run or those operational things are great and you can improve very much. But if you don’t add customer service  indicators or if you don’t include GE or a quality indicator that allows you to have a holistic view of where the benefit should be, then you start having benefits here and leakage here. And then the sum, and Ed mentioned it ends up zero or negative if you don’t  watch out. So  to having a holistic view, to having your operators, nowadays with the benefit of AI and all the data that can come even in the mobiles of every operator and they can have their top five indicators and check that, it’s a way to have everyone on the same view in a permanent way.

Ed Ballina (13:42)
Yeah, I mean, another one that we, ⁓ that I don’t think we’ve kind of taken advantage of yet, right? As you apply AI and predictive technologies to operating lines,  your efficiency goes up, your reliability goes up, right? Your equipment doesn’t fail without warning. But guess what? We probably are still carrying $15 million worth of storeroom parts that were sized based on a run to failure strategy, right? And the fact that you couldn’t get stuff from overseas overnight, right? Well, now if your equipment is equipped with AI, first, it’ll tell you there’s a problem. It will diagnose what it is you need to help to cure that problem. Perhaps it’s a new bearing. And if you take it all the way, it’ll order that bearing from the supplier to wind up in your shop a week before you need it. Now, if that’s the world we’re living in, folks, why do we need 15 bearings of the same thing worth a grand apiece sitting there?

Half of which have probably dried out grease on them and may not work. So that’s an opportunity, to, as we see these lines become more predictable, let’s fix the rest of our systems, right, to line up with that.

Alvaro Cuba (14:55)
The entire process, how much capital you have tied to spare parts. And now with predictive maintenance, you don’t need it. You can even share  spare parts among plants because you know that you are going to need this spare part in 15 days in this plant and not in the other. So you don’t need to duplicate.

So that’s the concept that automation is great and it gets a lot of benefits. AI is great, it gets a lot of benefits. But if you don’t do that holistically and you redesign the process, come on, as part of it, you will not get all the benefits. Probably you can get some of the benefits, but not all of them.

Ed Ballina (15:39)
Bingo. Agreed. Agreed. It’s like putting a new set of tires on your car, but never checking the tire pressure. Are you going to get something out of your new tires? Sure. But if you’re not running it at the right pressure, you’re not going to maximize life, traction, MPG, et cetera. Anyway, I think we beat that one to death. So now let’s talk about people that you and I always talk about, because there is leakage in people as well. So Alvaro.

Alvaro Cuba (16:15)
Yeah.

Ed Ballina (16:20)
You know, I know you’ve had some thoughts on this one.

Alvaro Cuba (16:22)
Yeah, and we talk a lot and Ed had a very good analogy that I’m going to steal from him, which is the lake analogy and the level in the lake. And when we start this journey, operators, mechanics, supervisors, everyone is running around like crazy. No? And bad decisions, not attention quality,  even personal problems. And then you start putting AI and you start putting automation and that start giving some relief and some breathing moments to the people, which is great. We want them calm, relaxed, time to think on safety, time to think on quality and.

And that’s the first level. And then when that happens, then you start seeing other rocks that you were not seeing before. So when you get into the next iteration, then you need to start thinking what else you can do with the people, no? So the first that comes is upskilling, reskilling, No? We have been talking about the need to have more skills, the need to have people that are multiple skills that can change from here to there.  With the new technology comes the need for new skills. And even now you can use the time of the people, the experts in the line to train the people that is coming on board. No, and to parent, parenting and everyone that comes in has certain amount of weeks where have a coach or kind of that to help. No, you can educate and that has a multiplying effect because as you train them in new skills and they start looking other things, they are the first to start coming back and say, now with this technology, we also can do this. And now we can take advantage of that. So, and that starts becoming a positive cycle.

Ed Ballina (19:03)
That  flywheel effect is incredibly powerful. And it could be negative or positive right? If you’re, if you happen to be in a plant where you’re you know running to failure and having staffing issues and bad morale that flywheel will take you negative right? But we on the positive side once you start getting some wins right and getting people involved and thinking about more than just, you know, what’s gonna break next? That positive flywheel effect will start manifesting itself in synergy throughout your lines. And it’s incredibly powerful. Look, we’re going to create, if you execute AI correctly, you’re going to create time for people, right? And nature abhors a vacuum, right? So if you create time, our suggestion is be mindful on what you choose to fill that time with. So if I’m an operator that is accustomed to my machine failing every five to seven minutes, I am running around like a chicken with my head caught off because it takes me two to three minutes to repair, you know, to, to mean time to repair. Right. If all of a sudden we get that line running where you have one shot, one, one 15 minute, I’m gonna say one five minutes stop every hour, you’ve just given that person time. What do you want to do with that? Right? For them. Right? So what does Mary Ann want to get training on? Maybe if they are a filler operator, they could spend time learning how the blow motor works so they can control more of her destiny, right? Or maybe if they’re running a paper machine while they’re in the control room, because things are stable, they can start studying for their next tech level, right? Or maybe they do a work with a maintenance technician so they can check off their maintenance skill block. It gives people a lot of opportunity to… Alvaro, to steal from you shamelessly, to maybe take them out to the stores and have them look at how their product looks on the store shelf, right? Or send them to a FAT. We were installing a line in one of our facilities many moons ago and I had the chance to go to Germany to watch the line run. And I brought one of my lead mechanics with me and my goodness, what, first of all, he was incredibly insightful.

And also he became such an advocate. owned that line. When that line started up, that was his line. It’s amazing what you can do when you reinvest the time and people get to grow. So fun stuff.

Alvaro Cuba (21:35)
I had an experience on that, but the mechanic and the operator of a powder beverage went to the EFT. And the first question was, how am I going to clean this? And these guys, they don’t operate, so they don’t know that.

And it generates a lot of dust and it’s very difficult to clean. So, immediately I said, no, this is going to be a terrible design, not work for us. And then working together, the EFT and our operators came with a design that was easy to clean, no? But that’s just an example. The other example, that I had the experience is you heard 71 % of the CEOs interested in  strong innovation and creation of new products. And I’m not sure if you guys are familiar with reverse engineering. I had this experience  in a chocolate line that was kind of particular chocolate line. And we were getting this asks from marketing and we were saying, well, it’s not possible. No, that’s not possible. And they were, come on guys, everything that comes from every ask you said, we can’t. And then we came with this we said, okay, guys. So we put together the mechanics, the quality operator and with artificial intelligence, you hit both. Remember, you improve your line capabilities, and then you can do even new products or different kinds of products. And at the same time, you are freeing time from your people. So now they can think, they can go to the shelf and see what the  competitors are doing. And they are the best ones to know their capability of their lines. And we start coming with a couple of products that we could run that we saw in other categories. And we came with that to the commercial area and the reaction was, oh my God, really? Can you do that? Absolutely. Six months later, we were launching a new product with the capability of the line, in a line that was 50 % utilized. So now you gain: One plus one plus one equals 10.

Ed Ballina (24:36)
That’s a great story.  I like you suffered  occasionally from being called the sales prevention vice president by my sales and marketing folks, because they would come to me with, hey, Ed, we have this idea here. And I was like, man, you know, this is going to cost like X amount and we can’t. And it was like, thank you very much, Mr. Sales Prevention. I was like, no, I’m trying to prevent you from eroding the P &L. But always, always had fun. No, those those engagements are really powerful and I guess the net takeaway from the conf… one other thing, one other factoid that really caught my eye– the, I think was BCG that you’ve quoted that 82 % of consultants are very very high on AI Okay, sorry, of course they’re going to be, that’s what they’re selling, okay? So, but contrast that to the CEOs who are like a little more timid and all joking aside folks, we all know AI is the way to go. And it’s, it’s, it’s a market that is not even in its baby steps yet. Right. So early adopters, you know, you take a chance, you’re not going to get it perfect, but you get in early. There’s going to be some missteps along the way. But I think if you focus on how you truly tie the work to a financial or some other stakeholder outcome and track that like Alvaro said with good KPIs, right? And then couple that with taking the time you’re freeing up from your workforce and engaging that time to improve their capability or their own well-being in the plant. You got to make sure you put that whole thing together because if you don’t, it has the potential to sub… to subperform. You’re not going to get the true potential.

Alvaro Cuba (26:32)
One example you would love that I experienced is  once we got all that and the machine was working much better and the operators have time and they were creating and all that, these couple ladies  had still a little bit more time and they said, well, what do we can do?

The plant manager said, we always  have been thinking on a training shop. No, and  they got in charge, put together the training shop and then new people start coming or people that were changing jobs and needing upskilling were coming to the training shop and they were teaching in their times. But you will even love more this one. When that was up and running and the shop was very nice and to the ultimate standards, one day the university in the city came and knocked the door and said, hey, we would love to use your training shop for our guys that are just about to graduate to go and experience in there.

Guess who was training these guys? The two ladies. Yes. So that’s the sum of the things because then you gain the benefit for the business and you also have a personal benefit for the people which allows you to gain engagement and that people is better prepared to take innovation and take other things to the plant. So could be great stories, but it’s very sad, on the other hand, when you go and see the productivity leakage, no? So automation and  AI high tech is great, but needs to be thought holistically as we thought, thinking about the people, the process. And the technology. And on top of that, the business and the customer needs. The customer needs, the business needs, and then you have all that because every business has one different strategy. No, you heard from the CEOs, some want growth, the others want productivity, the others want  innovation, the others want new products.

So you need to tune up with that and using these new acquired capabilities to get the business the edge that they need to win in the marketplace.

Ed Ballina (29:46)
Yeah, because unfortunately, team, if you don’t do that, there’s the potential that you could set back your AI evolution by years. Because if  you’re going to convince senior leadership to make a sizable investment in this technology and in training and support systems. And if it fails to deliver  there’s a chance that they will not be interested in making that kind of investment again in a number of years. So not only do you fail to deliver, but you set the cause back. Because now, you’ve got to convince people over again and if you didn’t deliver the first time,  your odds to deliver the second time  get a little shadier.

Alvaro Cuba (30:16)
Mm-hmm.

Great. Well, friends, that’s a wrap up for today’s episode. So thank you very much for joining us.  We are very happy to have you in the manufacturing meetup. And please bring your pals. Let’s make this meetup bigger. And  if you liked it, the episode, please follow us, subscribe us. ⁓

And ⁓ if you’re watching on YouTube, please give us a like or in iTunes, give us a review. More important, share it with the people that you know.

Ed Ballina (31:10)
Terrific. So as Alvaro said, let’s keep this conversation going. ⁓ One way to do that is to email us at mmu@augury.com or find us on the endpoint, a free online community for manufacturing pros like you and I and Alvaro, of course. ⁓ And that’s endpoint.augury.com. We’ll also have those links in the show notes for this episode. And also as you’re watching the podcast, you know, you don’t have to wait till the end to give us the like and subscribe. We won’t hold it against you if you are so enthralled that within three minutes you decide you really want to hear us again. So anyway, see you next time.

Alvaro Cuba (31:49)
See you guys, bye.