Transforming Integration: The Shift to Bridge Services – Antoine Shagoury, Kyndryl

This is Textron tv. Hey guys, thanks for the throw. We're here with Anton Shari, c t o, for ndl, and we're talking about how application integration is changing and maybe being converted into a surface with a little help from ai.

Anton, welcome the show. Thanks very much, Mike. Appreciate the opportunity Forever in a day.

We've spent, I don't know, countless hours integrating various things and using low-level programming tools to do that, and people had to stand up their own app servers and a million other things on top of that, and it's gotten tedious, cumbersome, and hard to maintain over the years. So how is all the way we think about integration gonna evolve? We hear a lot about AI these days.

I feel like we're on the cusp of something, but nobody knows quite what that is. Yeah, I, uh, I agree. We're definitely in a state of evolution, if I could use the word.

I don't know if it if it makes sense or not, but we're, we're, we're seeing great opportunities in how our customers are thinking about the integration and the approaching application, even data integration to get better into an opportunity for their business. And, um, one of the things that, uh, you know, whether it resonates for you or not, is we've, we're also seeing kind of a growing appetite for this type of work to move as a service. So that's really helping us or instigating us as a managed service partner to try to expand our libraries, our libraries of integration, data integration points, application, native integration tooling that we have even our own a p i opportunities to really make that opportunity easier for the customer.

Um, one of the other things that we also see, and I'm I'm sure you see this just as much as we do, is there's really a movement from a lot of the custom development and the custom integration work that the client's willing to take on just from a cost, time and risk perspective. So it's really shifting and uh, and one of the things I like to say internally is as much as we thought managed services was, you know, going away, it's not, it's evolving and it's evolving very much in this integration space. And, uh, yeah, I think we can tease it, and I'm sure you'll kind of add more questions to this.

AI is becoming a very strong component as we get better on data handling and data mapping and data categorization as we get to these companies and applications. Well, let's start with the fundamentals a little bit. We've been talking about build versus buy forever and always seems to me the history of integration was people built a lot of applications and then they said, by gosh, we need to integrate these things.

But are we kinda flipping the model maybe and starting with the integration platform and then wrapping the applications around it? I mean, have we been putting a cart before the horse all these years? Wow.

Um, little bit of leading to witness, but I actually like the thinking. It's, uh, it is definitely, we're now getting pace with how do we create the right tools and approach for integration. So it's building the framework before forcing, forcing the work to happen.

So to your point, we're going in head first this time on how we're approaching that, and that's where I think the, the efficiencies and opportunities really, really started to come to life for a lot of the clients moving into this modernization track with us In that dynamic, where will AI fit? I mean, am I one data just gonna walk in and say, I'd like to integrate these four data sources and have something magically occur, or, um, where's the heavy lifting gonna be? Well, I'm not sure if I should be depressed before I answer that question, or, or I think it's an opportunity there, but, um, and I'm sorry, it's, yeah, I have fun with that one too.

I, it might get there, it might get to the point where a lot of this work can literally be enabled through central natural interaction and natural processing and what you wanna do, kind of like the, the, the chat G t P enablement on how we do these things. But the, um, in initially I kind of see some really, so whether you believe into the, um, code or the code advisory capabilities, right, from some of the partners out there. So the co-pilot opportunities we're definitely getting better and providing better tooling to take some of the less custom work out of integration.

So we're trying to peel back that onion a little bit. Uh, if I were to go one step further, the more we can create these adapters, natural extension services, data integration services, they become very, I, I guess, open to using more artificial intelligence or augmented intelligence techniques to build those into the integration jobs that we need to do between applications, between business services and processes. So, uh, cutting back to the chase.

Yeah, I think maybe we may be out of a job, but I think, you know, I think we have a little while left to kind of help train, train the trainers and train the system to help take the, what I call the tedious work out of our job As part of that process. And I'm not sure that we'll all be replaced by a robot one day, but you never know. But it does seem to me like the functions and the jobs will change and evolve as the technology starts to mature, because today we have, yeah, programmers writing a lot of procedural code and then there's some low level people writing low, you know, using low-code tools to do the front end for that.

And then there's a data engineer and all these data ops processes, and then behind that as a software engineer with all the DevOps processes. And now we're thrown in some ML ops on top for the grin. Is there another way to think about the way all these things are structured?

Because it, at the end of the day, they are all forms of integration. Yeah, I agree. Um, you know, I really had, you know, I guess you're, the way you frame, it's great.

I, I think the, there is a new way to look at this and I think it is part of that paradigm shift and it's really breaking down how we can start to enable, and, uh, I, I use that word on purpose actually. How do we enable the developers regardless if they're developing, you know, into, uh, natural primitives or getting into low code or into application or functional code, how we create those natural libraries for 'em to use. So really creating an opportunity to improve their development experience.

I think there's some interesting things that can happen to start and how we create those. Um, and that's why you like the word augmented, but how do we create those augmented libraries to make the development experience not just easier, but more effective so we're not just wasting the time, we're being more effective in how we're building those integrations. And I, and I do love the way you characterize it, it is nothing more than what I call functional programmatic integration, right, at a lower level, right.

Toward delivering in the business. And, uh, and if, you know, it's funny, if you, if you take a little bit of that evolution in the market, you'll see many of us starting to adapt many of those different techniques as we've adapted more and more around microservices and event models, again, back into the operating infrastructure. So really, I think it's really complimentary and I, and I think there's a lot more embracing a lot more of an embracement, you know, opportunity with our developer community than I think most people realize.

It's really trying to accelerate the opportunity. You folks have launched your own integration platform that kind of drives down this path. But I wonder, are we, to a certain degree witnessing the democratization of integration?

Because a lot of this stuff was previously the purview of a global 2000 kind of company that could afford these resources, but maybe, you know, what we'll see is it'll become simpler, but more people will do it. Yeah, I, uh, oh, I'm not gonna cut to the cut right to the chase on this one. Yes, I think it is fundamentally democratizing, you know, how, you know, how we're not only develop the service, but how people consume it and how we're, and how they contribute to it, advance it.

And, um, outside of our integration platform approach, or even even with that in mind, it's really to create that medium, to develop that, to develop those capabilities, to create those opportunities for co-creation, for better tool contribution. I'd love we all, we've talked about open source crowdsourcing and community source for a while, these types of approaches that we're bringing to market really amplify that. So I think yeah, you're right.

I, i, it's, um, I had a conversation a few months back just to kind of reflect on something. You can laugh at me after I reflect on it a bit. You know, they said, you know, are you engineering to put yourself out of a job?

Almost like your first question that you opened up. And if you answered right, it's yes. 'cause you don't wanna stand still.

The current job you're doing now has to evolve. And the amount of complexity that we're seeing in the market through modernization, through the transformations that's actually happening, the compounding opportunities with data, look at the opportunities just being delivered just in the last six months on the focus around a opportunities and model development. Those opportunities really get, um, fueled by an open platform approach or a platform engineered approach to bring these solutions.

So yeah, I, I think there really is a democrat democratization on how we can not only build the solutions, but how we contribute to everyone's, I guess, use of those tools going forward. And I, and I, and I'm confident our positioning and now it's a little bit of us focus, uh, uh, we're positioning and helping our clients successful and making those transitions. And that's really why we're, we've investing in the platform models for our integration tools.

Part of using AI to accelerate integration is gonna be the building of the models, right? And whether you're using an L L M or machine learning or whatever it is, somebody has to build that model. And a lot of the organizations that I know that focus on integration or build their own platforms for that or do it internally themselves, they're not gonna have the resources to go build those AI models.

And if they had the resources to build AI models, they would probably build building AI models for something more interesting. So does AI kind of force the shift to more of a, an integration service mindset because somebody's gotta aggregate the data, pull all the code together and build all the AI models, and that's only gonna be the realm of a handful of companies? Yeah, I, um, I think it does make that shift.

I think it really starts to, and I, I often, I don't know if I say it as well as you do, but I kind of call it on demand or usage model or demand models and that specialization matters. And, and I guess where the difference is in that model development, it's the combination of yeah, the model, the training, but also the bottom underlying data and the use case you're actually optimizing against. So yeah, I, I think there's a, um, not only a tremendous shift, but a tremendous opportunity on how many of our providers and including us really position ourselves for not just that model development piece, but optimization, but also data harvesting.

So we're seeing the same challenges even just on normalizing and inquiring data, let alone building into that, that training case and building into the model, whether it's on a language model or not, or a statistical model. So we're seeing, again, a, a very strong growth, I, I wouldn't say it fully niche, but a growth in that on-demand capability or as a service capability for, for the customers Customers. And I don't wanna equate integration with roughly the it equivalent of, you know, taking out the trash, but it's not a fun job.

It's a lot of work in a lot of pain. And so maybe, you know, we're at the point where we could have people focus on other things that are a lot more interesting, like building application logic and fun stuff like that versus, you know, stitching all this stuff together all the time and then trying to babysit it for eternity. Oh, wow.

Yeah. I'm not sure how to react to that one, Mike, but the, um, the, you, you know, maybe I can recover my, but trying have a really sound like a smart answer back. The, um, you know, putting, you know, putting the right attention to what matters to the customer to the end result is the right way to shift the focus.

Yeah. And doing a lot of the integration work is busy work. I wouldn't say it's taking out trash, but it's def it's definitely busy work and the more we can improve the tooling, that's where I think a lot of the ai, if we can get into the copilot type of approach, a lot of the code abstraction away from that, yeah.

Focusing on human-centered opportunity, functional delivery of capabilities on the applications, I think it's gonna probably usher in even a newer form of better usability and better product, you know, positioning for the market. So yeah, I think there's, there's a, I think there's a good shift there, uh, if that, that sounded okay, but there's definitely a good shift in that space, although I'm not gonna get that taken out the trash thing outta my head for a while. Alright.

But seriously, what is your best advice to folks about, there's big changes coming. We can see them, we don't know exactly how they're gonna manifest themselves, but how do I prep for that now? Do I, you know, put everybody in a room and have a strategic plan?

Or do I just kind of start playing with stuff and see where it goes? Well, I, I wish there was a perfect answer to a direction, but I, I, I'll probably echo back what we think is a good approach. So we've started with a lot of, and this gonna sound almost fundamental, but where are the skills?

Where's the work happening within the application development space and how are we building the right tools? So how do we get the right, so how do we start to look at the teams that we can move into that functional development space and that user development space within that? And how do we then have the other teams start to look at what are the things at which are, what are the tools, the extension services, the APIs that we can really create an easier environment to do a very low code approach or no code approach to a lot of the fundamental integration points across the applications, the data tooling, the databases, um, as well as the business services.

So we've started down that path and, and, and, and it works well. The, the toughest thing beyond that is the plan. Where are we, where do you want to develop into, it's not just having people break down their work.

Hey, I'm doing the job for the job. I control this SAP integration point, this interface, this data, this the basic data transformation layer. No, no, no.

I need you to actually look at what the end result is now. So going back to the old development model, how do I understand what the user needs, how they, the experience they want to operate it in? And that's how we've started to approach it.

So I think there's a, I think there's an approach. We went skills first. We went into segmentation around those skills and getting the teams at can very much focused on that user-centered design.

And then have the other teams start to train how we build the tooling, the templates, and the integrations. And even we are playing with the combination of automation and augmented intelligence and how we're bringing the code together. So it's a really interesting approach to it.

We, we think that works and it works well. All right, folks, you heard here, it may be relatively simple. You come up with a list of things you like to do and things you don't like to do, and then figure out where integration falls into that bucket and have the machine do the part you don't like to do.

Hey, Anton, thanks for being on the show. Absolute pleasure. I'll bring my magic wand next time too, Mike.

Thanks So much.