IFS Unleashed: How Digital Workers Are Becoming Digital Teams

[00:00:00] - [Speaker 0]
It's not your fault that your agentic AI systems are acting outside of compliance. There were just too many datasets to guardrail them all. But with Denodo, you can now organize your hundreds of data sources into one layer and govern your agents with a single approach. Try it now with Denodo by visiting denodo.com to learn Welcome back to the Tech Talks Daily Podcast, and I'm recording this episode today from IFS Unleashed twenty twenty six in Orlando, where one of the biggest conversations is about what happens when AI stops answering questions and starts getting work done. Now if you're a regular listener, you may remember that last year in New York, I spoke with Somya Kapoor, CEO of IFS Loops.

[00:00:57] - [Speaker 0]
And last year, we talked about the arrival of individual digital workers. Fast forward twelve months later, those workers are now being coordinated in digital teams, handling everything from procurement and maintenance to scheduling and customer service. But how much autonomy should we be giving these systems? And who steps in when something goes wrong, and what happens to the people that are working alongside them? Well, with customers reporting significant efficiency gains, including Kodiak Gas reducing certain parts related tasks from thirty minutes to just a couple.

[00:01:32] - [Speaker 0]
And we're moving beyond theory into real business outcomes here. So I'm delighted to say that Somi is gonna be joining me again today to share what organizations are learning, where the unexpected challenges might appear, and why managing change could be proving harder than implementing the technology itself. But enough for me. Let me beam your ears directly to the show floor here right now. We're recording live here at Unleashed twenty six in Orlando, and the last time we spoke was New York last year, and the the conversation was about individual digital workers.

[00:02:09] - [Speaker 0]
But this year, we're talking about digital teams. So what has changed? What what can

[00:02:14] - [Speaker 1]
these teams do that a single digital worker maybe couldn't have done while we spoke last year?

[00:02:18] - [Speaker 2]
Yeah. You know what? The single digital worker was very much involved in a microprocessor and a small outcome that can be derivative. The reason that we learned from an experience from the market is where the projects were failing because you were trying to boil a big ocean and kinda taking up a bigger process. And, you know, as you heard Pedro and Lance kinda talk about it, those projects are never ending projects.

[00:02:39] - [Speaker 2]
They never finish. After one year, you kinda go figure out, oh my god. I'm still at the starting point. So digital worker was our way of articulating to the market. You know, start with a very focused, low risk, high impact workflows.

[00:02:52] - [Speaker 2]
And within IFS, we identified some of those last year, and that's a starting point to show quick validation. We had most of our customers go live in less than twelve weeks. Now it's getting down to four weeks of out of the box digital workers, and then we just kept building. And that building is a compounding effect that you get with a digital crew to a point now we successfully can say customers at least that started six months ago are live with the two, if not three, digital workers, and they're expanding more and more across it. So now we're getting down to procure to pay, asset life cycle management, first time fix, quicker SLAs.

[00:03:31] - [Speaker 2]
Right? And for that, we recommend three or four digital workers in terms of that are working hand in hand.

[00:03:37] - [Speaker 1]
And one of the great things about coming to conferences like this is there's so many different conversations and use cases. But one of the things I love about this conference here, we're not talking about the future. We're talking about right now. You don't have to name many names, but are there any real world use cases of how you're seeing digital workers or teams of digital workers and the kind of impacts that they're having?

[00:03:54] - [Speaker 2]
No. Absolutely. I mean, we shared 10 examples of the customers that are seeing efficiency gains. Right? Lance itself from KLM is seeing close to 60% efficiency gain across one, and he's deployed the second one, which is the customer order manager as well.

[00:04:09] - [Speaker 2]
We have got Moe from Eight Group that took in a knowledge worker and job intake manager and is seeing close to... Now if that runs annually for a year, he's got 30,000 invoices per month. He's going to see... And each invoices cost them roughly about, you know, 5 to £6. And if they're running like this around the whole year, with the time savings, even 80%, if we were able to process them, he's able to see close to 1,000,000 in pounds savings.

[00:04:41] - [Speaker 2]
Right? And that gives him... Me... Ability to get the next digital worker, the next digital worker. You know, Kodiak gas itself has seen close to 87% efficiency gain from the, material replenisher.

[00:04:52] - [Speaker 2]
Now this is a unique customer... Sorry, unique use case because, you know, when we talk about part search, it doesn't do part search just. It actually does searching of the parts, within sites, then outside sites from vendors, and then it actually goes and puts a purchase request in depending if the... If if the part was available on-site, then it goes and tells the technician, okay. You know what?

[00:05:16] - [Speaker 2]
Go and get it from this site. But if it's not, it puts a purchase request in from a vendor that it needs to be sourced from as well. So it runs downstream automation, once it's able to identify the parts from a natural search, standpoint. Right? So that itself, typically, within this environment, can take about anywhere from fifteen to thirty minutes, and that's brought down to a minute or two, 87% efficiency gain across his organization on that.

[00:05:44] - [Speaker 2]
And that's giving him an ability to moving into material sourcing, into AP invoicing, where the compounding effects of all these digital workers are gonna be a lot more.

[00:05:52] - [Speaker 1]
And I suspect we will have people watching and listening that are very early on in their journey, and the idea of trusting a digital worker with certain tasks might be a little bit intimidating. But you'd have to do everything at once. So where would you recommend that that they start? What kind of tasks should they be given first, and what are they comfortable doing? The the customers you're speaking with, what are they not so comfortable doing?

[00:06:13] - [Speaker 2]
See, within within our digital worker, this is exactly right. Our customers, you know, everybody else in the market is giving you, here's my agent studio. Go build everything. Right? And soon enough, you realize it's actually technology that's not the barrier.

[00:06:27] - [Speaker 2]
It's the change management of it. So what we have done with the digital workers is kinda giving you a predefined template if you wanna look at efficiency gains within field service process or, you know, giving you templates around asset management if you wanna see efficiency gains over there or procurement or finance related. And these are the areas where you can kinda pick and choose where you wanna get started. We give you, okay, an area. Do you want to, you know, use an agent to do entry of a work order or a service request or a procurement purchase?

[00:06:58] - [Speaker 2]
And then downstream, you know, then do inventory replenisher or material replenisher or service close out manager. So each one of these is additive in nature, but, you know, we we kinda gave you a road map where to start and then kinda keep adding and layering more and more to drive more efficiency and productivity gains.

[00:07:17] - [Speaker 1]
And worst case scenario, let's say an agent delivered an unexpected outcome or something that was incorrect, where does accountability sit there?

[00:07:25] - [Speaker 2]
So, you know, every digital worker comes with a super... Supervisor agent, so there is a complete audit trail and approval process. If the digital worker, like any intern in your environment, is making a mistake, you can go self correct it from... By giving feedback to the agent itself, and it self learns and self evolves. That's why having these digital worker agents are very important in your environment because they're learning for every interaction and every input that you give, and they're self improving the process within your environment, if not identifying gaps that might be existing that you probably didn't know before.

[00:07:59] - [Speaker 1]
100%. And are there any real world examples you can share where digital workers have maybe changed or evolved the roles of of human employees as well?

[00:08:08] - [Speaker 2]
Actually, you know, most of our customers are saying that it's became... Bringing behavior changes. Ah. Right? People are people are liking to...

[00:08:16] - [Speaker 2]
They start with one. They get comfortable. They're getting used to and driving outcomes efficiency. For for instance, I can search for parts very easily in house and then put a purchase request in, which kinda took me fifteen to twenty days. My customer order manager are automatically converting into work orders.

[00:08:34] - [Speaker 2]
That's increasing my customer experience and my interaction with my customers. I'm able to do inventory replenishment, understand where the parts are, which site has it, and then kinda put them collectively in. You know, job intake manager, converting my thousands of requests coming in into proper work orders and then using dispatcher assist downstream to really get the... It's scheduled and, you know, fed into a technician, and sent on-site is reducing my SLAs down. So every area, we have customers where KLM is seeing 60% efficiency gain.

[00:09:08] - [Speaker 2]
Moe is seeing one thousand three hundred hours saved per month, across one site, and there are multiple sites. You know, Pedro is seeing 90,000 hours saved annually. You know, We have we have multitude of different customers giving, you know, numbers of efficiency gain, number of hours saved, with these digital workers in pretty much in weeks of deploying them.

[00:09:32] - [Speaker 1]
And within weeks of deploying them, how many of them, when they get further on down the journey, have real unexpected results? Do you hear hear stories of things like that that they maybe set off, wanted to achieve something, and and unlock even more yet?

[00:09:45] - [Speaker 2]
Do they You know, actually, we have another customer, that that has got to 85% completely no touch, and he wants to get to a 100%. Woah. And he's... This 50% that he has, 15%, he keeps saying, oh, but that vendor did not... Like, you know, this vendor did not recognize because they were missing five five things.

[00:10:05] - [Speaker 2]
And we we kinda say that's a long tail problem now

[00:10:07] - [Speaker 1]
Yeah.

[00:10:08] - [Speaker 2]
Yeah. Because, you know, you have to keep feeding it. And the next time it comes from the same vendor, then, you know, the agent recognizes it and does it. But if it's gonna be a new vendor that is not in the system, or if it's an invoice coming that an agent does... Has not seen before, it's definitely gonna seed into a human in the loop over there.

[00:10:26] - [Speaker 2]
And that's the edge cases I feel like are coming now. Right? So... But that's a long tail problem, and it'll learn and it'll evolve on how to interact with those as well.

[00:10:35] - [Speaker 1]
And when people look at the challenges, how much of it is AI, and how much comes down to fragmented data, legacy systems, unclear processes, or or just organizational readiness and company culture?

[00:10:46] - [Speaker 2]
You know, I think the the thing about, this technology from historical... I mean, I've been in the enterprise space to tell you enough that every, you know, decade, we had, you know, achieved digital offers or digital boardrooms, and everybody needed the cleanse data, and you have to bring data together. With this with with this agentic technology, right, what it's gonna do is it's gonna find gaps within your data, your process, your people understanding. And I think you have to look at it differently in terms of how to evolve it. It does have limitations.

[00:11:17] - [Speaker 2]
If you... If there's bad data and you're seeding bad data into a system like this, it is gonna give you a bad output. But what historically you didn't know what what was the bad data and where I need to fix it, now seeding into a digital worker, you'll be like, oh, you know, this vendor always sends me an email with no vendor name. And if you don't have a vendor name, I can't upload it into IFS. Right?

[00:11:38] - [Speaker 2]
So what do I do? Those are the things that you feed into the system as a feedback, and then that's how it learns and it evolves.

[00:11:44] - [Speaker 1]
Well, as I said, we'll come full circle now. The very beginning of our conversation, we're talking about one year ago, it was digital workers. Now it's team of digital workers. Any predictions or anything that excites you for what our conversation could be like in another twelve months? How do you see this evolving?

[00:11:58] - [Speaker 2]
You know, actually, you heard Pedro say that. He wants to come back where he's deployed hundreds of agents on this stuff as well. I mean, I was like, maybe not a 100 agents, but probably another 10 more digital workers that could be deployed using, another 100 agentic skills. But, you know, we are here. We're evolving.

[00:12:15] - [Speaker 2]
I think we're gonna be bridging a gap between the digital and the physical world in a year's time, and it's gonna be an exciting time. It's moving too fast.

[00:12:22] - [Speaker 1]
Awesome. Well, a huge thank you as always for joining me. Hopefully, we'll join you again next year, but thanks, everyone.

[00:12:28] - [Speaker 2]
Yeah. Same here. Thank you. Cheers.

[00:12:31] - [Speaker 0]
Wow. Massive thank you again to Somya Kapoor for joining me today this time from IFS Unleashed in Orlando. What I found particularly interesting about our conversation is just how quickly the discussion has moved forward. Twelve months ago, we're talking about individual digital workers completing specific tasks. One year later, we're seeing teams of AI agents working across entire business processes with some organizations already reporting substantial time savings.

[00:13:04] - [Speaker 0]
And another big standout there from our conversation was that acknowledgment that technology isn't necessarily the biggest obstacle. Very often, it's managing change, dealing with incomplete data, handling unexpected exceptions, and knowing when humans need to step in. And get that right. I think her example proves volumes there from a customer achieving 85% touchless processing. I think that reminds us that the remaining 15% could be where some of those hardest problems continue to hide.

[00:13:37] - [Speaker 0]
So I'll be sharing more conversations and lessons from IFS Unleashed. Please visit techtalksnetwork.com if you wanna speak with me. And as always, big, big thank you to each and every one of you for listening and listening to the end. I'll be back again tomorrow with another guest. Thanks for listening.

[00:13:54] - [Speaker 0]
Bye for now.