'Systems of record must change' - IFS Ultimo CEO Steven Elsham shares his agentic AI plans for EAM

In my Enterprise and Misses weekly, I regularly assess the the AI bubble question - and how gen AI projects have gone off track. These reports are concerning, and they are piling up. 

However, most of the big studies are really about the underwhelming results of gen AI for generalized tasks. Or: trying to 'wing it' with your own gen AI projects, without the aid of specialized firms (and the specialized architectures they can provide). 

One of my core contentions: we'll see better results from enterprise AI when domain experts embed AI in industry settings, and bring (secure!) customer data into these scenarios. 

The tech fantasy happy notion that an employee can just grab an LLM like an order of fries and do effective things at any type of scale is fraught with problems.

So where does that leave us? If we want more impactful enterprise AI, we need a better architecture - and yes, better proof points. At IFS Connect in June, I documented that story: Industrial AI meets AI agents, but how are customers deriving value? Now I have the chance to see how IFS Ultimo stacks up, via a virtual discussion with new Ultimo CEO Steven Elsham. 

Can IFS Ultimo change the market with "best-of-breed," agentic EAM? 

Like its parent company IFS, Ultimo is geared towards asset-intensive industries. IFS Ultimo bills itself as "asset management EAM software focused on maintenance & safety with a fast time to value, short implementation time, and ease of use." For Ultimo, that means cloud-based EAM, running on Microsoft Azure. 

Elsham joined as IFS Ultimo CEO in May 2025. Which prompts the question: why Ultimo? As Elsham told me, he is drinking the "digital labor Kool Aid," but not to replace SaaS, or the human talent that uses it: 

A few years ago, I drank the Kool Aid on the rise of the digital labor market and the potential in the digital labor market.  I could see from the deployments we were doing with some early clients that we were working with very closely how digital labor could augment what the traditional highly-skilled human labor could do in various areas. I was particularly focused on our marketing and commerce products at the time.

 I could very quickly see how the deployment of that digital labor could radically improve the outcomes that our clients were able to deliver with our standard SaaS software. So I didn't see it as replacing SaaS software. I saw it as an augmentation of SaaS.

Elsham could see the potential to inject the same sensibility into cloud EAM - and he told IFS the same. 

When I was approached by Ultimo and the IFS team, and they talked to me about Ultimo and what they were looking to do, I told them what I would be doing if I was running the business. We had a very strong confluence of interest there. What I saw was Ultimo was generally recognized as one of the category leaders in doing what it does really well, which is a best-of-breed EAM system.

That best-of-breed aspect is important: Elsham believes Ultimo can outperform the EAM [Enterprise Asset Management] provided by broader ERP solutions.

In maintenance operations, you are typically serving the production function of a business. You are avoiding downtime. You are trying to make sure that those production processes are as efficient and effective as possible. My view was: taken in the right way, we can materially affect those core KPIs.

Mix that with agentic AI, and Elsham sees a big moment for Ultimo. He believes that agentic AI can address a "significant failing" in the enterprise software industry: to figure out what's going on, you have to go into the software. For plant managers, that's from ideal. But can we change that? 

The way enterprise software works at the moment - which is a significant failing of our industry - is that we expect people to come to us. We produce the software, and we expect people to come to us and log into the software.

I'm old enough to have worked in environments where there wasn't much software. In that environment, we used to interact with human beings who used to ask us questions, and presented us with decisions and information... I think we can recreate software using that kind of metaphor - so we can go to where the human is.

The agentic shop floor update - "Have I got time to brush my teeth?"

Elsham cites the example of that super-fun early morning alarm clock. Wouldn't plant managers welcome a pro-active update, on the state of their equipment assets? 

If I'm a plant manager, what I really want is: I want to know as quickly as possible, 'Is everything okay?' There's been a night shift, is everything all right?' I don't really want to log into an app and look at something, and scroll down and see some dashboards and see how many incidents were logged by the night shift. I just want to know: 'Is everything okay? Have I got time to brush my teeth, get in the shower, talk to my wife, feed the cat, grab a cup of coffee before I leave the house?' The best way to do that would be for an agent to send me a message just saying, 'Everything's okay with the night shift last night.' And then I know I can do those kind of things.

And no, logging into Microsoft Teams doesn't cut it either:

I don't want that as a Teams message. I'm not logged into Teams at the moment; I'm not looking at Teams. I want that as a rich message. I want that as an iMessage, because I'm an iPhone user... I haven't actually arrived at my desk yet. 

Elsham takes this scenario further: 

[Let's say that] last night, there were two minor incidents with a breakdown of a conveyor belt, but they were addressed by the night shift operators. But if we want to address that, so there isn't a problem with the day shift, we're going to need to bring in two technicians to fix that during the day shift. Those two people will need to come in on overtime. 

'The cost will be X, do you approve? Yes or no.' Yes. 'I don't want to go into the office, go to my desk, log into the system, go through two factor authentication, smile at the camera at the top of my laptop, go through all that process, to look through a series of charts and see two incidents were logged.

A pro-active system would change that: 

I don't want to ring somebody up and say, 'Hey, how serious is it? What's the best way of addressing it? Okay, can you bring those two engineers in?' I want that to happen automatically., I want the system to be reaching out to those engineers, to make contact with them, to have a phone call to ask them: 'Do you want to come in for overtime?'

Bottom line: "systems of record" need to change. 

We need to, I believe, change the way that we interact with software  - to make the system of record more functional. 

So how is Ultimo doing on that mission? Elsham says it's full speed ahead, staring with EHS [Environment, Health and Safety]:

We released the first of those agents about a month ago, which is our EHS Manager. It's starting off relatively simple. Eventually, I think we're going to be providing digital labor. But the simple thing it's doing now is looking through work orders - and it's looking for things that could be an incident that should be reported. 

But does EHS Manager fit into the human operator's workflow? 

If we identify that this door is being damaged in some way, and where it's been damaged, we can probably presume that it's been probably hit by a forklift truck, because we've had a number of those incidents in the past. If it was hit by a forklift truck, then that should be reported as an EHS incident. 

So we're doing that incident reporting, and then submitting those incidents through to the human operator, for the human operator to say, 'Yes, I agree.' We can go in and modify some elements of it, but 'Yes, I agree this should be reported.' Now, this isn't saying humans shouldn't be reporting these things when they occur. What we're saying is we've got a safety net now.

My take - on digital labor, headcount reductions, and getting agentic AI right

Elsham's ambitions for IFS Ultimo don't stop with AI: he also wants to drive expansion in the US - and fuel global growth with timely acquisitions. To that end, Ultimo is fresh off its August 2025 acquisition of FSI, a US-based player in healthcare facilities and technology management software (This acquisition is Ultimo’s second in the US, and its third overall since 2024). 

I'm not generally as excited about acquisitions as vendors are, but I do like specialized industry moves like these. As per IFS, "FSI has the only computerized maintenance management system (CMMS) platform purpose-built for healthcare service professionals." That's the kind of context today's AI needs. 

I never seem to have enough space for the spirited AI debates I have at the end of these interviews. And there is plenty to debate on how to get AI agents right, and where they can go wrong. Some executives don't agree with my takes on this, but that doesn't matter. What I want to see is the openness to discussion - because customers have plenty of questions also. 

Though I'm not a fan of the "digital labor" catchphrase, I didn't take that up with Elsham. Why? Because in the industries IFS Ultimo services, most customers need all the human talent - and automation - they can get. If you want to call it machine talent, I'm not going to argue - as long as it helps teams solve real problems.

If I'm a customer, I want a vendor like IFS Ultimo to give me as much automation as possible, but with plenty of controls on how much autonomy I 'activate.' 

I like the early use cases Elsham described, where agents play quality assurance roles. With the right data context, that type of Q/A agent can give human workers something they haven't had before. Customer education plays a big role here. If a customer told me their AI agent says "everything went okay on the night shift," I would encourage that customer to ask the agent follow up questions, just to make sure that the agent is pulling from the right status reports and information (LLM agents are not strong on sequential time or causality, though there are ways to improve that in a specialized context). But Elsham's right: we've never had this kind of interaction before. Why wouldn't we want that? 

When equipment assets go down, money and time quickly falls into that sinkhole. Getting out in front of that, while providing pro-active information to production teams - that's a big win. To pull this off, we'll need a strong collaboration between vendor and customer, not to mention a willingness to digitize shop floor assets, and upgrade legacy plant management tools. Otherwise, the necessary data will elude us.

In the next ten years, a significant portion of the skilled labor in industrial settings will retire, taking invaluable data for AI systems with them, via personal experience. How do we make the most of that labor now? How can we be capturing what they do, in a format AI systems can use? Elsham's team is pursuing those questions, and they are the right ones:

We can address part of the labor shortage by taking all of the cohorts behind those most experienced workers and bringing them up to a level of experience that wouldn't have been achievable in that timescale before, because we're eliminating the non-value-added stuff. If you actually look at where somebody is actually truly learning and developing expertise, it's a relatively small proportion of the working week. How do we accelerate that? 

That's where Ultimo's customer pilots come in: 

We're working really hard on that at the moment, and working out: how do we ensure the right balance between the density of data capture, the quality of the data capture, but also making that as efficient and effective as possible... We were quite lucky. We've had customers for quite some years in Ultimo who are very, very helpful in helping us try and explore these things.

The big AI vendors seem to roll out 'workplace productivity' solutions nobody asked for, and then raise prices. But If Elsham's team builds AI systems out of these EAM field projects, I like their chances a heck of a lot better.