Podcast: How IFS Loops is redefining Industrial AI for high-volume operations
Key Highlights
- Agentic AI workers help manufacturers automate high-volume tasks and improve operational efficiency.
- AI pilot failures stem mostly from change-management challenges, not the technology itself.
- Digital workers reduce technician downtime by enabling faster, data-driven decisions in the field.
- AI platforms create new industrial skill sets, helping manufacturers close the growing labor gap.
Industrial AI software heavyweight IFS chose its New York show to debut virtual agents equipped with 50 skills that autonomously manage high-volume tasks. In this episode of Great Question: A Manufacturing Podcast, Smart Industry's Scott Achelpohl and Plant Services' Thomas Wilk were there to chat about the product, IFS Loops, with its CEO, Somya Kapoor.
Below is an excerpt from the podcast:
SA: Hello everyone, and welcome to another great episode of Great Question: The Manufacturing Podcast, brought to you by two Endeavor B2B brands, Smart Industry and Plant Services. I'm Scott Achelpohl, Head of Content for Smart Industry, and I'm joined by my colleague, Tom Wilk, who is Editor in Chief for Plant Services.
We're coming to you from the IFS Industrial X Unleashed Conference in New York, and our guest on the pod today is Somya Kapoor. She's the CEO of IFS Loops, which we found out today is a very interesting and forward-reaching product. It’s an update—if I may say an update—of the company’s AI agentic platform. IFS debuted an update at the conference, what it calls the next evolution of Loops here at IFX Industrial X. So welcome to the program, Somya.
SK: Very excited to be here, folks.
SA: IFS Loops is a product that delivers industrial customers templated digital workers equipped with 50 agentic skills now, today, as we speak—right? And supposedly 100-plus by next month, December 2025, right?
SK: That is right.
SA: Somya, today during your presentation, you told us that the pace of operational change is faster than ever, with complexity growing faster than human teams or legacy systems can handle. Operations leaders and CIOs can’t wait for months of IT rollout to benefit from new capabilities. IFS Loops gives companies a digital workforce they can deploy today. Tell us more about your product.
SK: IFS Loops is an industrial-grade agentic platform that we have from the acquisition IFS did in June, and we already have customers deploying it in production. What we’ve done is given everybody this agentic capability that helps you build and redefine your processes by mashing data from different systems, where you’re not constantly bogged down by, “Oh my God, my data is in one system and the other,” but really rethink a process from a pain-point and business-outcome standpoint.
For instance, a Supplier Order Manager really helps you process your orders autonomously—sitting in your inbox day in, day out. Right now, you might have two people manning that inbox, five people—some of our customers do—but they’re doing that manual job of taking and looking at every PO entry, then taking that PDF or email body, finding a part ID, and manually loading it into IFS Cloud.
Well, now the digital worker, the Supplier Order Manager, can not only upload that data automatically into IFS Cloud if you’d like, it can also communicate with your customer if certain parts of that order are missing—like shipping information, part information. It also looks at the part conversions within IFS Cloud and makes the conversion necessary so you can successfully update that within the IFS Cloud data.
Now, we do understand AI is not perfect, so there are always exceptions. So how you do exception handling is also something we do as part of the agentic platform. So it’s not just about deploying these digital workers, but also: how do you monitor them? How do you create an audit trail? When handling exceptions—just like a human on the job—how do you retrain them? All those factors are provided out-of-the-box today.
TW: It’s different than just tracking down Joe or Jane on the floor who put the wrong lube in the bearing, right? It's a whole new frontier when it comes to addressing AI traceability.
SK: Exactly. And that’s what is the key aspect of industrial-grade solutions, right? Because these are deployed in highly regulated environments, and you want to make sure that there is an audit trail for everything that the AI is doing. So as part of the platform, we do have a supervisor platform—a supervisor agent—for every digital worker that keeps the audit trail and monitors. And we leave it up to you to do the change-management aspect of it, rather than just saying, “Oh, this is coming from a plain automation,” which is very static in nature.
TW: In your relationship with IFS—I know it’s fairly longstanding—but the formal acquisition took place earlier this year, like June or July, correct?
SK: That was June, yes.
TW: How did your companies get to know each other better, and then what led to the acquisition? How has that been going?
SK: IFS was a customer of Loops before the acquisition. They actually bought us in the CX space as an agentic solution to help optimize their support operations. Soon enough, they started digging deeper, and they’re like, “Oh, we’re fishing for some agentic solution in the market,” and they had conversations, and this seemed like the perfect marriage—because we were moving out from the CX operations into other operations that we wanted to get into. And because of the platform approach that we had taken, where we were not just a platform to build agents but also to monitor, test, deploy, version, and handle security and governance—that really attracted them to us, and that’s where the marriage happened.
Now, how is it going? It’s going actually really well. We’ve just been, what, 120 days in? We have a few customers live in production in the industrial space, using our digital workers that we launched out-of-the-box, and actually scaling to everything else. So it’s quite fascinating. And we did have our first customer on stage who actually articulated the value—what was it? They went live in seven weeks using Material Replenisher, and that is getting them $3 million in cost savings and 90,000 hours saved.
SA: That's a lot of money. It sounds like you're in the honeymoon period to me.
TW: Well, the savings were located in the time saved by the technicians who didn’t have to look for the correct parts, right? Yep. At this point, it always seems like the first part of ROI that’s found is with the frontline employees who aren’t slowed down anymore by either surprise interruptions to the production line or time spent looking for parts. I mean, did this surprise you at all, or was this exactly where you and the customers looked to generate savings?
SK: The reason why we picked these digital workers to begin with also is, you know, it depends on the process where you can show the buck for the money fast, right? Because you’ve got to bring people along. This is not here to replace human beings—I want to be clear on that—but it is to really amplify the ability of what you can scale and do a lot more with less.
So when Pedro came and said, “This material replenisher is a big pain point. Most of my technicians know the parts, but it’s actually at that 3 a.m. in the night when they can’t physically call a supply chain manager to get the part ID, and he has to wait or she has to wait for four hours to make that happen. It’s a huge loss of money for me. So how can I create a frictionless operational experience for my field technicians, who are doing the hardest job, the most dirty job, and make life easy for them?”
So that kind of drove this digital worker for Pedro to a point that now he wants to jump into the customer order, into the supplier order, into inventory replenisher—and each one of them are building blocks for the next operational step. And at the end of the day, unlike our FPA systems, these agents will talk to each other. So you're creating a network of them within your environment for operational efficiency.
TW: One of the points that came up multiple times today was the secondary impact of AI agents helping to remove emotion from the work equation. And what I mean by that is, you mentioned the 3 a.m. call that someone’s got to take. Without the AI agent in place, you're going to get an emotional reaction from that worker at 3 in the morning. Could be angry, could be resigned—could be a combination of all that. But these agents are helping change the workforce into something less driven by emotional reactions and more into analyzing what is the best path forward, correct?
SK: Exactly—into frictionless experiences, I like to call it. And that’s very true, right? Some of the manual tasks and repetitive tasks—why do you need a human to do that when you can use humans for the best cognitive ability that they have, where agents have not caught up to that yet, right? So I think that is where the aspect comes in of using these reasoning, self-evolving agents. If I’m missing shipping information right now for a supplier, I will remember it the next time once you correct me. So that’s the thing that they do provide. And you know what—they’re 24/7 on, and they don’t need health benefits.
SA: You don’t have to pay them for overtime.
SK: That’s right.
TW: That’s interesting too, because when you put a display of all the different roles that were possible with IFS Loops, one of the questions that Scott and I chatted about was: does this represent the jobs that are being lost, or the jobs that are being covered for that can’t be filled? And as a bit of background on that question, I think it is the latter, and it’s jobs that are increasingly difficult to fill.
I was at a conference earlier this year where we took a look at the percent of reactive work versus proactive work being done by maintenance teams, and how that number hasn’t really shifted in about 10 years—even though the labor force has drawn down over the past 10 years. But what’s been covering that gap? Automation—tools like this. So I’m curious to know: is that part of the motivating factor of the IFS move, to fill in the gap as needed?
SK: It is, actually. Like we said—boosting your productivity, upskilling folks, and finding folks for skills where you just can’t find people. It’s all of those elements, right? In that process, you might also realize that you’ve over-hired, so then it’s actually driving productivity off of those things as well.
So I think it’s in all elements—it’s bringing efficiency within your environment. And I tell all my customers that this is real. You know, I’m based out of Silicon Valley, so I’ve drank the Kool-Aid already, but I like to tell people that this is unlike the technology… I know I’m not giving my age away, but I’ve been in now three decades of evolution of technology really happening for the last 30 years. This is very real.
This is the era where you relate it to the internet coming in, right? And that fundamentally will bring jobs in. So, you know, we all have to learn how to now monitor these agents—that’s a skill set within ourselves. Managers don’t understand how to do that stuff, right? Because they’re so used to doing performance reviews for humans. How do you do performance reviews for an agent? It’s a new skill to build, but it’s a reality that your job will require that moving forward. So the earlier you get started, the more competitive advantage you have.
SA: As Tom pointed out, as a filler of the skills gap, I’m going to note—and we’ve noted it in several stories recently—that, what was it, Deloitte who estimated that by 2033 we’re going to need 3.8 million more workers worldwide in manufacturing and industry? And they estimated that 1.9 million of those could go unfilled because the skills just aren’t there to fill them. AI will have a definite role in helping to close that gap.
SK: AI will have more creation of jobs. As you heard Deloitte and MIT also highlight, it’s creating net-new opportunities and net-new roles that we need more sophisticated skills for, and we need people to start ramping up on that today—because it’s a huge gap.
TW: My publication covers lights-out, fully automated factories, and we also cover the small amount of shop operations that have duct tape on the conveyor belts. What percentage of plants do you think are ready, infrastructure-wise, for this revolution? I mean, are we past 50%, 60%? Is it sort of a have-and-have-not situation?
SK: You know, I don't have the numbers at the top of my mind, my head, but what I would say is that if you're not ready, you won't have the competitive advantage. You're going to miss out on that. And I don't think it's about having the data in the perfect format. It's not about having the systems in the perfect format. It's about having this technology in place to tell you what the gaps are there—what operational efficiency you need to drive. And if you're not able to, what more you need to bring in to get there. Because if your competition gets there first, you're going to be standing behind.
SA: Somya, something that was kind of shocking that came up today was— is there really a 95% failure rate in AI pilots so far? I noticed that the 5% was emphasized, but 95%… what's causing the failure rate?
SK: You know, I've been in this space now and having my software deployed within the support arena, and kind of seeing the benefits of it, it's not the technology that is leading to the failure. It's actually the change management and the mindset shift that you've got to bring up. People are scared, they're fearful, so they reject the technology—like, “Oh, it doesn't work, it's not accurate.” But you don't bring the same level of accuracy when you're hiring someone to do the same job. You expect AI to be 100% accurate on day one, whereas your employees are barely 20% accurate on day one, right?
SA: That failure, that fear that AI is going to take my job.
SK: I think that fear, and the change management of how to approach it, is leading to those failures. It's not the technology. And I'm not saying this because I'm in it—I've been living this for the last four years. And I've seen it truly make an impact for people who are using it the right way.
TW: I hear you. One of the convincers that I've seen in my field is that if you develop an AI agent that, number one, is optional— for example, on people's phones for field workers— and number two, if those field workers are prompted to enter more information, that helps, like guided questions. For example, if you've got someone in the field who needs to report back what happened to a motor, and the technician says, “Motor burned out,” the agent can easily say, “What else did you observe?” And then more information— well, skills are there. And so it's iterative, and after a while, the fact is just getting in the habit of simply being more open with their initial diagnosis of the situation.
SK: Yep, and there is another one— we were talking to a customer where they said, “Can you bring”— and we showed the demo too— “can you bring upsell opportunities?” Right? All of a sudden now you're converting your service into sales motions as well. So while I'm there fixing one part— by the way, they just licensed some other one too— can we do that as well? So those are the things that are getting to the next level, opportunity-wise, which were never there. Now you can have your field technician saying, “Oh, wait, I'm here to fix this part, but actually I got notified there's another part that needs a warranty update.” Or, “I'm here, I need to get this updated because I can upsell you this one too. What about that one?” So the things that were not done before can be made happen right now.
TW: And for the humans on board, it's broadening skill sets too.
SK: Exactly.
SA: Well, Somya and Tom, I think we've got to wrap it up. Great stuff— great explanation about what Loops does. It's fascinating. And with that, I can't thank you enough, and Tom enough, for joining us on this episode of Great Question: A Manufacturing Podcast. From Smart Industry and our friends over at Plant Services and our friends at IFS, we say goodbye to our listeners and say have a great rest of your day.
About the Author
Scott Achelpohl
Scott Achelpohl is the managing editor of Smart Industry. He has spent stints in business-to-business journalism covering U.S. trucking and transportation for FleetOwner, a sister website and magazine of SI’s at Endeavor Business Media, and branches of the U.S. military for Navy League of the United States. He's a graduate of the University of Kansas and the William Allen White School of Journalism with many years of media experience inside and outside B2B journalism.