SEI taps IBM for agentic AI growth

Transcript - SEI taps IBM for agentic AI growth

Beth Pariseau: From Informa TechTarget, I'm Beth Pariseau, and this is IT Ops Query.

This podcast distills the signal from the noise about enterprise software development and platform engineering. Each week, we'll talk to expert guests about the latest tech industry news and trends that engineering and IT leaders need to know.

Don't forget to subscribe to IT Ops Query for more conversations on AI and the future of the enterprise digital workspace.

Greetings from Boston. I'm here at IBM Think 2026 with Zach Womack, CTO responsible for SEI's technology strategy, execution and delivery for all SEI platforms and applications globally, including the SEI wealth platform Trust 3000 and investment management products.

But that's not all. Zach is also responsible for enterprise architecture, cloud and data strategy, design, development, delivery, DevSecOps and production operations for the financial services company. He has spent more than 20 years in various roles at SEI.

In March, SEI engaged with IBM Consulting to conduct a review of its operational systems and workflows, aimed at advancing agentic AI automation.

So, thank you very much for taking the time out.

Zach Womack: Absolutely. That was a mouthful, but it's a pleasure to be here.

Pariseau: Yeah, I mean, you have just a few hats that you wear.

Womack: And now AI has become one of the main hats I have to wear. So, yes, a lot of hats.

Pariseau: So, I guess, maybe just kind of set the stage as far as what you're using AI for currently. You know, do you have AI agents running around? What does the estate look like at the moment?

Womack: I would say that we kind of have a broad spectrum running. One of the ways that SEI looks at this is that, you know, we don't want to have some sort of separate AI strategy. AI is something that's going to be fundamental to our business. So realistically, our business strategy has to be AI-conscious from the bottom to the top.

You know, three years ago, we started on this sort of journey around AI. You know, three years ago, we started our center of excellence, we started doing a lot of PoCs and so on and so forth, but that has matured tremendously. I would say every individual at SEI currently is AI-enabled in some way.

So, you know, we've got things like M365 Copilot for, sort of, the average employee. But we also have, sort of, specific applications that we use depending on the role. So, our development community has a number of, like, code-assisting products and so forth. So, at the individual level, we've deployed a lot of tooling.

When you think about SEI, one of our core competencies is technology and technology product. I mean, that's one of the things that we deliver to the market. And a big part of technology product now is AI. So, we actually have some capabilities deployed in production in our products. That's something that's relatively new.

A lot of what we've done from an agent and AI capability are things that are being deployed and consumed internally first, because we always kind of want to have that human in the loop. But that's obviously changing and maturing as we go along, and we have to figure out where do we, kind of, put these capabilities directly in front of our clients.

Pariseau: Can you give me an example of where you have AI agents in production, like what kinds of workloads they're automating?

Womack: An example would be, like, in client service. So, we're bringing in a lot of information related to previous tickets and service history, along with a wealth of knowledge associated with a lot of our user guides and release notes and things like that.

So, within our SWP application, which services our wealth business, our client service community now has a capability to basically answer questions effectively, in real time, with high-quality information because we have an agent that was built to read and pull all of that information in.

And that allows the client service personnel to still have a direct line to the client that's asking the question, but they're able to answer those questions faster and better than they were previously -- with the eye towards, this is the type of capability that we want to roll out directly to our clients.

Pariseau: Right. And so, is that also using IBM AI tools already, or is that new?

Womack: Currently not. IBM, from an AI tools perspective, is relatively new to us.

Pariseau: OK.

Womack: We engaged IBM really around our operational needs. We did a really thorough RFP process when it came to this, actually.

And looking at IBM, they really stood out as a thought leader here. And they brought some proprietary tech in that really gave us the idea that, like, as we engage with them, they're going to be able to really analyze and figure out from a workflow perspective not just where do you apply AI, but how do you do this most effectively and really deliver on the promise that AI holds -- as opposed to just coming in and automating small pieces of a workflow -- but really reengineering the workflow from top down and bottom up.

Pariseau: So, what technology was it that kind of got your attention that they brought in?

Womack: From a pure tech perspective, some of the things that they have is, really, the tools that their analysts are going to be utilizing.

Pariseau: Oh, OK.

Womack: So, it's the technology that IBM is going to use. The other thing that I think is interesting is -- it's maybe not a one specific piece of tech. It is the open platform that they provide. Because one of the things that I worry about as a leader in the space is vendor lock-in.

So, one of the concerns I had in looking at this is having somebody that's going to come in and recommend, let's say, just all of their products, and then you're going to be stuck with that. You know, once you bake vendor products in, it's very hard to get them out.

So, the fact that they're able to come in, look at what we have, what we're missing. They might be able to supply some things in the area that we're missing but leverage the capabilities that we already have in-house. That was very desirable to me as well.

Pariseau: OK. So, do you plan to deploy any of IBM's AI tools, or is it more about optimizing what you're already using?

Womack: I would say chiefly, for me, it's optimizing what we're already using. But obviously, this engagement has, you know, turned me on to various things and capabilities that they do have.

I mean, obviously, they're touting Bob here today. I think that's a really intriguing platform. The fact that they're kind of looking at full-stack -- or maybe full-stack is not the right word. I would say, sort of, beginning to end, the SDLC process. It's something that we're actively looking at right now today, which is, you know, how do I reengineer the SDLC process in light of AI and how rapidly things move?

If you think about the way development has worked historically, the long pole in everything was the development time. Right? Like that development took the most time, so everything kind of got built around that.

Now, as we think about reimagining the SDLC, you know, the longer threads aren't necessarily the development time frames. It's, like, how quickly can you build out prototypes, and how do you automate testing and all those kinds of things. And what I've seen from Bob, so far, I think that's, you know, a compelling platform, I will say, that we should be evaluating as we look at, you know, building out on our SDLC.

Pariseau: OK. You mentioned Microsoft 365. What about GitHub Copilot? Is that in place?

Womack: That's in place as well, yeah. We use a variety of tools on our development side. I think that, again, looking from a vendor lock-in perspective, I think what's important to me is that we apply the right tool to the right situation.

So, depending on what the stack is that the developer is working on, you know, the relevancy for a particular tool set could change depending on what type of development they're doing -- whether they're sort of in the back end in the database or whether they're working in the front end -- that modifies, like, potentially the tool set that you want to use.

One of the things I think is important to say here: We really need to continue to develop out our framework. When I say that at SEI we've enabled the individual, I think we've done a really good job at getting in front of this and giving people, at the individual, the tools that they need, sort of, in this modern world and modern economy.

I think the reality for us is, you know, we need to do that at the enterprise level. And what I mean by that is if I've got an operations person that has built an agent that they're leveraging, I want them to be able to share that, and I want that to go through a governance model. I want that to be deployed in some type of marketplace where people can -- other operations users could potentially leverage that.

It almost becomes like a CI/CD pipeline for nontechnical users, which has never really been needed or been put in place before. And I would say the same thing for my development staff. You know, I've got a lot of really good things happening where people are really becoming more agent orchestrators than they are developers. It's really changing the nature of what development means.

But again, that's happening at the individual level for us, and we have to figure out how to scale that across the organization.

Pariseau: So, it sounds like you might be looking at something like Watsonx Orchestrate?

Womack: Certainly.

Pariseau: OK. So, the overall goal, like you say, is to scale across the enterprise. Can you give me an example of, you know, a workflow you'd really like to transform with AI that, you know, might be one of the initial targets?

Womack: Well, I think that there's a lot of aspects of our operation that we look at and say there's a lot of friction.

So, I think the classic one in wealth industry is the client onboarding process. So, if you look at that wealth, that whole process, it's very human-intensive today. So, that makes it an expensive and long process for SEI to do. It's also a kind of cumbersome and difficult process for the client.

So, as we look at, you know, onboarding a client, going through AML and KYC and all of that, that's a prime area where we could look at AI as an ability to remove the friction from that process.

Pariseau: So aside from, you know, the ability to be a bit vendor-agnostic, at least as far as hybrid cloud goes, and work with different tools, was there anything else about IBM, you know, expertise that made them kind of stand out in the RFP?

Womack: Yeah, I think, you know, it's interesting that, you know, now we're talking about AI and machines doing everything. And I think that the reality is people really still matter. I think that's true for SEI. I think that's one of the things that I think will really serve us well into the future, which is we have long-tenured, highly experienced, highly knowledgeable people that are knowledgeable in our systems as well as the industry.

And if you look at that, what AI will do is allow us to scale that. Right? So that's the SEI side of it. I think we saw similar with IBM. We see their sort of pedigree and being able to come in and look at operational systems and do systems engineering and operations engineering around that. And I think that's something that we think that they're going to be able to really excel at for us.

Pariseau: So, do any of their other foundational things -- like IBM Concert platform or Watsonx.Data -- either of those something you are going to look into?

Womack: Certainly, I think that now that we're engaged with IBM -- because IBM, historically, for us has been more of like a hardware provider -- as we're engaging with them and kind of taking them up from a partnership perspective, I'm certain that we will be looking at all of those things.

I think what I previously mentioned are probably the ones that are right in front of me right now. I think that's what we're focused on. But yeah, I think that, you know, I'm looking for opportunity wherever it lies,

Pariseau: Right. Does that include mainframes, the hardware?

Womack: Yeah. Yes.

Pariseau: OK, that makes sense. Right. And so, what is the biggest challenge you envision, you know, as you undertake this review and look to scale AI agents? What do you think is gonna be the biggest?

Womack: Oh, there's so many challenges. But I think governance is one. I think that things are moving very quickly, as everyone says.

But I also think that we're really starting to, you know, within our organization -- I'm sure other organizations are experiencing this -- we're getting to a point where the opportunity set and what we're able to execute against that is speeding up. And the governance model around that is our governance model. And I think most are highly, sort of, human-centric and process-driven. And that's really become the pinch point for us.

And that's something that we're looking at in terms of what can we do as governance as code? You know, things like that, where we are really focused on making sure that the appropriate governance is in place because that's, like, an imperative right now. But I think that efficiency is going to be also important, or you're going to start to lose some of the efficacy of the governance itself.

So that's one. And then I think I'd come back to what I said before, which is being able to scale what an individual does across the enterprise. Again, I don't think that that's really something that most organizations have done outside of, sort of, their technology organizations. And maybe I'm a little bit biased on that, coming from the technology side.

But, you know, like I said, this is starting to become sort of an 'operator as developer and orchestrator of agents.' Well, that requires different tooling and different skill sets that aren't really there yet.

Pariseau: Yeah, how is it going to change the nature of work and how people get work done? What kinds of skills are necessary, and how do you address that within your organization?

Womack: I know, well, it's almost certainly going to change the nature of work. I can say on what we're already seeing on the technology side is sort of the 'orchestrator of agent' processes, right?

So, again, I think having people that are really knowledgeable about the context of what you're doing and able to create and orchestrate agents that will execute tasks is something that I'm already seeing within the technology space. So, my best developers are operating like that today.

Pariseau: OK.

Womack: And I think when you look towards operations, it's going to become sort of a similar skill set. People are going to have to fundamentally still know the nuts and bolts of what it is they do from an operational perspective, day in and day out. But ultimately, they're going to be managing agents that are doing it on their behalf. So that changes the nature of work.

Pariseau: So, what's the result of shifting into the orchestrator of agents?

Womack: I think the benefit to the business is speed and quality. I think there's a lot -- and scale, right? So, I think that there's a lot made of the 'Oh, expense reduction' side of this. But I think ultimately that's a little bit of a race to the bottom.

And what I mean by that is if we can provide our services, you know, from a scale perspective, that means that we can do and sell more business, right? We also know quality is king, right? In the financial industry, it's got to be, right? So, to the extent that it improves quality and speed, because speed in a lot of cases is highly valuable to our clients. Those are the three things where I think it creates a better overall capability and service that we can deliver to the market.

Pariseau: What do you do to ensure quality? Because you hear so many horror stories about AI agents that just go off the rails or they make things up. I mean, how do you ensure that?

Womack: Well, one is human on or in the loop. That's one of the things that we really focus on. So, the work product is the individual's, not the agent's. Right? So, that could change over time.

One of the other things that we're looking at is competitive AI. So, in areas where you have two different models or two different processes that arrive at the same conclusion, you can be highly confident in that conclusion. If they don't, then it goes through an exception process. So, that's kind of two examples of models that we're following to ensure quality.

Pariseau: Doesn't having that -- models judging models -- doesn't that increase the cost as well, operationally?

Womack: Certainly, I mean, you're touching on another difficult subject, which is FinOps associated with AI. I mean, that's a huge challenge, you know, it's something that we had to deal with in the cloud world. You know, that's been, I don't want to say 'sorted out,' but people really understand that much more now and it's something that can be managed. I mean, here today, I see multiple vendors at the conference talking about 'We can give you visibility into your token usage' and so forth.

And then, you know, model management, how do you decide which model you're using, and all of these kinds of things are top of mind right now, quite frankly, as we are, you know, continuing to develop out our 'AI harness', for lack of a better term, that we're going to be using across the organization. Because, you know, as individuals become agent creators, who defines the ROI on that? You know? So, like, these are some questions that still aren't necessarily answered but will be in time.

Pariseau: So, imagine if you will, IBM Think 2027. What would, ideally, be where you'd like to be by then, as far as this effort with IBM and AI agents?

Womack: Well, I certainly think we should be in a spot where that harness that I talked about, SEI has built and deployed that, and that's being effectively utilized. And then I think that, on the IBM side, is they've helped us define and build and deploy the agents that are enabling our operational workforce.

So, if I'm sitting here a year from now and the combination of those two things has been successful, that means that we've significantly improved our operation over those 12 months, and I think that would be a success.

Pariseau: Right. Operational workforce, as opposed to the technical folks that were the early adopters?

Womack: Exactly. Different tool sets but, sort of, merging disciplines when it comes to what you're actually doing.

Pariseau: Right, right. Wow.

Womack: Yeah.

Pariseau: OK. Well, just a few things with your many hats to worry about.

Womack: That's right.

Pariseau: So, yeah, I mean, is there anything else that you've heard at the show that caught your interest or that, you know, is on your mind right now?

Womack: Well, I got to see Andre Agassi earlier, so that was fun.

Pariseau: Oh yeah, that's really cool.

Womack: Yeah, no, it was a great keynote. And, you know, I think we're seeing how impactful this is across all industries. Like, I sit within finance. So, you know, sometimes I've got the blinders on for that. But hearing people get up there and talk about, you know, the potential benefits of the combination of AI and quantum computing has on healthcare -- that, you know, that's very interesting.

Pariseau: That's far-out stuff.

Womack: Absolutely, yeah.

Pariseau: Cool. Well, thank you so much for taking the time.

Womack: Absolutely.

Pariseau: Really appreciate it.

Womack: Pleasure meeting you.

Pariseau: Good to talk to you, and thanks very much for you out there for watching.

Thank you for tuning in to IT Ops Query. To learn more about enterprise software development and platform engineering, explore our content on Informa TechTarget sites. Find us on YouTube at our channel, Eye on Tech. Subscribe to our podcast to receive the latest episodes as they drop. And if you liked what you heard today, give us a rating and review on Apple, Spotify or wherever you're listening. Thank you for joining us.