2894: IBM and the AI Revolution: Enhancing Enterprise Applications

[00:00:00] As your organization unlocked the full potential of its hybrid cloud strategy, and have you

[00:00:07] done that through application modernization? Now many of you might not know this but before

[00:00:13] I was a podcaster in tech right out, I frequent at the world of IT and I spent a lot of time

[00:00:18] in application management. I still get flashbacks of trying to get 200 applications ready for

[00:00:24] Windows 7 and some of them were legacy apps that got no chance of working in the Windows

[00:00:29] 7 environment but that was a long time ago. In today's episode I sit down with an incredible

[00:00:36] guy, you know when you meet someone and you just hit it off with him straight away and

[00:00:39] you could just talk with him for hours? Well Kyle, he's one of those guys, he's

[00:00:44] an IBM Fellow and CTO for IBM Z Software, an all round great guy. And together today

[00:00:52] we're going to dive into how Generative AI is set to transform application modernization

[00:00:59] and why it's essential for every IT leader and indeed business leader so buckle up and

[00:01:04] hold on tight as I beam your ears all the way to California where Kyle is going to explain

[00:01:10] how this technology is not just enhancing legacy systems but also reshaping the future

[00:01:15] of enterprise workloads. So a massive warm welcome to the show Kyle, can you tell

[00:01:21] everyone listening a little about who you are and what you do? Sure, yeah, I'm an IBM

[00:01:27] Fellow and a CTO for Z Software organization. A lot of my focus is really and has been for

[00:01:33] some time now on application modernization and really hybrid cloud strategy and integration

[00:01:37] so a lot of cool opportunities come along with both of those areas so that's largely

[00:01:42] what takes a lot of my time up these days. Fantastic, it's a pleasure to have you

[00:01:47] join me today. Every day I try and demystify a different topic here on the

[00:01:53] podcast and with the emphasis on IT spending here in 2024, a lot of businesses

[00:01:59] challenge we're doing more with less is also that problem of technical debt

[00:02:02] and everybody's leaning towards modernizing applications at the same time.

[00:02:05] Can you elaborate on why enterprises should maybe be prioritizing this and

[00:02:10] also how it aligns with creating a more tailored intentional strategy for

[00:02:15] enterprise workloads? So much going on out there at the moment, isn't it?

[00:02:18] On both sides of the pod. No, there really are. At the end of the day, right,

[00:02:23] there's sort of two options I think that clients can choose to take and one of

[00:02:28] them is really terrible in my estimation which is let's modernize by migrating,

[00:02:34] rewriting, let's redo what we've done for the past several decades.

[00:02:39] There's a lot of assets that have been curated on the platform for quite

[00:02:42] some time. Clients have these assets, right? And so let's enhance these assets,

[00:02:47] right? Let's modernize these assets and I think that's the important bit but I

[00:02:51] will say that there are clients that feel that there's a little bit of a risk

[00:02:55] potentially in doing this in general, right? And so they do like weird things

[00:02:59] like, you know, let's do updates to our kind of our core fabric that

[00:03:03] really backs our business. Let's do updates like maybe once a quarter,

[00:03:07] right? And they think, well, we're really minimizing risk by doing so.

[00:03:10] But the reality is it's just the opposite, right? Because they're

[00:03:12] introducing a quarter's worth of change all in one go and hoping it all

[00:03:17] just stitches together well. That's really not the right way to do this here.

[00:03:22] Introducing a ton of risk by doing that obviously. And so what we have now

[00:03:26] is the ability for JNAI to really kind of help not take the reins.

[00:03:30] So that's not what that's not what our what we think AI should do is

[00:03:33] take the reins but instead really assist these developers and assist

[00:03:36] the modernization practices moving forward to really help future prove

[00:03:39] the enterprise. You mentioned technical debt, right? So how do we

[00:03:42] future prove the enterprise, right? And bring in for example, kind of

[00:03:46] potentially in some cases, the languages that our clients really

[00:03:50] need to be strategic for their platform for their future like

[00:03:52] let's help bring those new languages in, let's deliver a better

[00:03:55] experience not only for the external consumers but also

[00:03:58] their own developers like let's make that experience better for

[00:04:01] them, which is going to go straight to agility. Let's make it

[00:04:04] so much easier for their business to respond to their own

[00:04:07] business needs, their market demands. So a lot kind of gets

[00:04:10] wrapped into modernization but it's all good for the business and

[00:04:13] we think JNAI can really help accelerate this modernization

[00:04:16] play.

[00:04:18] And I do suspect for many people listening they wouldn't have

[00:04:20] immediately thought about or even linked to Generative AI and

[00:04:24] application modernization and the role of Generative AI in

[00:04:28] application modernization is it in fact pivotal? So can you

[00:04:32] explain how developers could maybe better leverage Generative

[00:04:35] AI for code translation and modernization? And what makes

[00:04:38] IBM's approach unique in this area?

[00:04:41] Oh, no, certainly. I think we've all seen sort of

[00:04:44] JNAI still relatively nascent in this space. And I think

[00:04:47] largely what we've seen a lot is the ability to go from like

[00:04:51] natural language to code. We can do things like, Hey, please

[00:04:56] generate me an insertion sorts in Python or in COBOL or in

[00:05:01] Java or whatever language you might desire. Like that's

[00:05:04] sort of where it kind of started in the JNAF for code

[00:05:06] space, but we're really getting much more sophisticated now in

[00:05:09] actually doing language to language like programming

[00:05:12] language to programming language and having JNAI really help

[00:05:15] accelerate that space. And so that's really largely what we're

[00:05:19] focused in IBM Z on our application modernization initiative,

[00:05:22] right, which is about how do we take, for example, COBOL as

[00:05:26] a programming language like Enterprise COBOL not kind of

[00:05:29] textbook COBOL but Enterprise COBOL. And how do we help

[00:05:32] leverage AI to actually transform that COBOL to another

[00:05:36] target language. In our case, we're targeting Java as our

[00:05:39] first initial kind of four into the space. And it's not to

[00:05:43] say that COBOL is not strategic for a platform

[00:05:45] because it is, but also so is Java also so was node so was

[00:05:50] go so was Python. So all these languages are also

[00:05:53] strategic and for those clients that really wants Java to

[00:05:58] be their strategic enterprise language of choice, we are

[00:06:01] helping them do that transformation from COBOL to that

[00:06:03] language. And I will add, it's what you asked what makes our

[00:06:08] solution unique in this space. It's a full lifecycle approach

[00:06:11] just with AI sort of as the heartbeat. But there's a whole

[00:06:14] I guess body built around that that that heartbeat. And

[00:06:19] that's the lifecycle that we provide. So for starters, it's

[00:06:22] about application understanding, right? It's about

[00:06:24] actually really refactoring helping to decompose these

[00:06:28] applications into discrete business services. Then we

[00:06:31] actually use a mix, we think our sweet spot is a mix of rules

[00:06:35] and AI. Traditionally, it's all been a rules based approach,

[00:06:38] but we're bringing some rules but then really leaning on gen AI

[00:06:41] to really help make this really this transformation like look

[00:06:44] really good and perform very well, and not be that

[00:06:47] traditional kind of Joe ball, where it's just kind of

[00:06:50] COBOL syntax expressed in Java, which really frankly hurts

[00:06:53] me. And it hurts our clients that have tried to go down

[00:06:57] that path. It's really not recognizable as job, it's not

[00:07:00] really maintainable. You're not hitting that goal of future

[00:07:03] proofing and agility. So we have kind of all that baked in and

[00:07:06] we do this incrementally as well. It's not big bang. It's

[00:07:09] really one business service at a time. And we know IBM knows

[00:07:15] right how to run mixed language workload on our

[00:07:17] platform incredibly efficiently, all within the same

[00:07:20] transaction that'll work. So COBOL and Java, even

[00:07:23] assembler, all working together in the same

[00:07:26] transaction to work. We've been doing this for a long time

[00:07:29] because frankly, like Java's legacy to we use legacy as a

[00:07:32] this kind of a bad word sometimes but Java has been on

[00:07:35] our platform for 25 years. We've been investing in that

[00:07:38] space for a long time. This is not a new space for us to

[00:07:40] get into. And so really that that whole like life cycle

[00:07:44] that we have is what makes our approach unique. And what I

[00:07:46] didn't mention was we also auto generate tests to validate

[00:07:50] what we've done throughout the process. And we're just

[00:07:52] trying to roll some of that out in this last

[00:07:54] order, but there's a lot more to come. So that's a really

[00:07:56] long answer to a I think a short question, but there's

[00:07:59] just so much to our solution that we're really proud

[00:08:01] about that really does make it unique in the space.

[00:08:04] Absolutely. And if we also throw in economic uncertainty,

[00:08:08] global conflict, businesses challenge with doing more

[00:08:11] with less and budgets continuously shifting. I've

[00:08:14] gone to ask what are your expectations for evolving

[00:08:19] ways in which generative AI will continue to support

[00:08:21] enterprises on these modernization journeys? Because

[00:08:24] a lot of businesses need a little help. A lot uncertain

[00:08:28] about generative AI. There was a certain amount of

[00:08:30] anxiousness last year. I think that is changing now.

[00:08:33] But what are you seeing here? Yeah, you brought up

[00:08:36] sort of like this the anxiousness around it in

[00:08:39] your right because a lot of it was how do we

[00:08:41] validate? How do we trust? How do we trust what

[00:08:44] this did was the right thing? And so I think as

[00:08:46] we're in bed again, I kind of go back to technical

[00:08:48] that like our clients, many of them too many of

[00:08:51] them, I should say not all right, but many of them

[00:08:54] actually don't necessarily have really good strong

[00:08:58] muscle around kind of automated testing. And so

[00:09:02] every time they make changes in their applications,

[00:09:05] it kind of really in their estimation like is a

[00:09:09] worrisome thing to do which again is counter to

[00:09:12] what we want to do in this industry, right? What

[00:09:14] we want to do with the innovation we can do

[00:09:15] with software development engineering. And so where

[00:09:18] we think gen AI is can also really help here this

[00:09:21] year and in the next is around that test case

[00:09:24] generation and that synthetic data generation in

[00:09:26] support of these tests. And I'll also say, I think

[00:09:29] we're going to see it later this year, most

[00:09:31] certainly going into next year, we're going to

[00:09:33] see going beyond just language code generation

[00:09:36] right, programmatic language code generation

[00:09:38] right, we're also looking at modernizing the

[00:09:40] way automation has done the platforms as well.

[00:09:42] And again, this is another big area of focus

[00:09:45] for a lot of our clients, right? How do we

[00:09:47] bring in sort of more contemporary, more

[00:09:49] modernized automation frameworks and have them

[00:09:51] support the Z platform, right? And so that

[00:09:54] automation is going to be really significant

[00:09:55] in the context of Z. What do you have, right? Our

[00:09:58] clients have a lot of JCL. They have a lot of

[00:10:00] Rex, right? These are how they do both sort

[00:10:03] of automation and scripting on the platform.

[00:10:06] Well, we're actually going to be leaning on gen AI

[00:10:08] to do a transformation of modernization of

[00:10:11] that into Python for scripting into

[00:10:14] Ansible for sort of that broader sort of

[00:10:17] bringing in the whole automation control plane

[00:10:20] in support of Z. We've had a lot of investment

[00:10:22] in Ansible over the years, specific ZOS

[00:10:24] collections curated out on Galaxy and

[00:10:26] Automation Hub. And our ability to help

[00:10:28] our clients modernize their existing

[00:10:30] automation versus saying, hey, you got to

[00:10:32] toss all this away and start fresh here

[00:10:34] because this is a great place to be. No,

[00:10:36] that's awful, right? For net new use cases,

[00:10:39] sure. But they have decades of automation

[00:10:42] they built up and curated. We want to

[00:10:43] help them modernize that to Python and

[00:10:45] Ansible. And we think that's a big space

[00:10:47] for our clients moving forward. And

[00:10:49] lastly, I'll also say this might come up

[00:10:52] again later, but explanations a big

[00:10:54] element, right? That's not necessary

[00:10:55] about switching languages, but better

[00:10:57] understand what's there today. Because

[00:10:59] a lot of our clients do find Cobalt

[00:11:01] strategic again, so do we, right? But

[00:11:04] they want to explain those assets that

[00:11:05] they've had for 10, 20, 30, 40 years

[00:11:07] so they can then better understand

[00:11:09] where the different moving parts are

[00:11:11] so they can better understand how to

[00:11:12] enhance and modernize and improve on

[00:11:14] that Cobalt and add new business

[00:11:15] services to their Cobalt applications

[00:11:17] as well. So explanation, I think is

[00:11:19] going to become a key enabler for

[00:11:21] our clients moving forward as well in

[00:11:23] the coming years.

[00:11:24] And before you came on the podcast

[00:11:26] today, I was reading that the average

[00:11:28] small business with 500 or fewer

[00:11:30] employees has something like 172

[00:11:33] applications at mid market companies

[00:11:36] between 502 and half thousand

[00:11:38] employees have 255 apps on average.

[00:11:41] And that number continuously increases.

[00:11:43] And of course, they're scattered everywhere

[00:11:45] now there's no longer a server in a room

[00:11:48] nobody quite knows what it does, but

[00:11:49] it does something for the last 10 years.

[00:11:51] It could be very complex managing

[00:11:54] application estates now. So modernization

[00:11:57] is key to execute executing

[00:11:59] a successful hybrid cloud strategy.

[00:12:01] So how are you at IBM envisioning

[00:12:04] the integration, this integration?

[00:12:06] How are you imagining this evolving

[00:12:08] and what are the primary benefits for

[00:12:10] businesses in adopting this approach?

[00:12:12] Because we can't just carry on going on that way.

[00:12:15] No, you're absolutely right.

[00:12:17] And I'll say even in IBM, there was a time

[00:12:21] where public cloud was the destination.

[00:12:24] And while we talked about hybrid, it was

[00:12:26] really just sort of viewed as a rest stop

[00:12:28] like along the way to this nirvana

[00:12:30] that was public cloud.

[00:12:32] And the market has proven out that's

[00:12:34] really not the reality.

[00:12:35] Right. Hybrid actually is the

[00:12:37] destination for a number of reasons.

[00:12:39] Right. Some of which we covered earlier

[00:12:40] about, hey, you have all these assets,

[00:12:42] modernize these assets if that's not

[00:12:44] if that's the right thing to modernize

[00:12:46] these assets and integrate them

[00:12:48] with the new assets that you're creating.

[00:12:49] Because as you just stated, long gone

[00:12:52] are the days where the entirety of an

[00:12:53] application is deployed to a single

[00:12:54] platform. Right. That has been

[00:12:56] gone for quite some time.

[00:12:57] Right. The entirety of an application

[00:12:59] is consuming a number of

[00:13:00] a number of public cloud services,

[00:13:02] private cloud services on-prem

[00:13:04] business and data services.

[00:13:05] That's what it's all about.

[00:13:06] Right. Every platform out there just

[00:13:08] provides an amount of business

[00:13:10] and data services.

[00:13:11] And you need to consume those services

[00:13:12] very easily and very cost effectively.

[00:13:14] And so that's what we're really helping

[00:13:16] to achieve.

[00:13:17] And it's also not

[00:13:19] it's not a it's not a binary

[00:13:21] proposition here.

[00:13:22] Like is it our Republic cloud?

[00:13:23] Are we on-prem, private cloud?

[00:13:25] Like it's it used to be viewed

[00:13:27] as binary. Right.

[00:13:28] But it's really an and value

[00:13:29] proposition. All these things coming

[00:13:31] together.

[00:13:32] Right. So it's really take advantage

[00:13:33] of what you have, add to it,

[00:13:35] integrate with it.

[00:13:36] And frankly, we're not all

[00:13:38] fighting over pieces the same darn

[00:13:39] pie. Right. This is not a zero-sum

[00:13:41] game in this space. As you just

[00:13:42] mentioned, there are new startups

[00:13:44] popping up all the time.

[00:13:47] Be our own enterprises are

[00:13:48] building new experiences all

[00:13:50] the time. There's so much more

[00:13:52] innovation and more experiences the

[00:13:53] world's going to deliver in this

[00:13:54] space that required new development

[00:13:56] to actually build these

[00:13:58] experiences that we can't

[00:14:00] spend our time just rewriting

[00:14:02] stuff that's already there.

[00:14:03] Right. Spend our time building

[00:14:05] the new experiences, adding

[00:14:07] the new innovation to the market

[00:14:08] and then integrating all the

[00:14:09] different things that you already

[00:14:10] have together in one cohesive

[00:14:12] kind of deliverable that we can

[00:14:14] share with our clients, whether

[00:14:15] they're internal clients or

[00:14:16] external clients. So I think that

[00:14:18] the fact that we're like

[00:14:19] settling on hybrid as indeed

[00:14:21] the destination just unlocked

[00:14:22] so much opportunity for us moving

[00:14:24] forward to really innovate at

[00:14:25] scale. But again, making sure

[00:14:27] that all of these complex

[00:14:28] working parts of an application

[00:14:30] are all really brought together

[00:14:31] really seamlessly.

[00:14:33] And I'm curious from the

[00:14:35] conversations you're having

[00:14:36] with businesses of all sizes

[00:14:37] around the world, what are the

[00:14:38] most significant challenges

[00:14:40] enterprises are facing in

[00:14:41] application modernization?

[00:14:43] And how are you IBM helping

[00:14:45] overcome some of those hurdles?

[00:14:47] I would imagine you've heard

[00:14:48] and seen so many different

[00:14:50] stories throughout the world

[00:14:52] and different business sizes.

[00:14:53] Right. Yeah, Neil, you're not

[00:14:55] wrong. I would say one

[00:14:57] of if not the

[00:14:58] the biggest challenge a lot

[00:15:00] of our clients have frankly is

[00:15:02] application understanding.

[00:15:04] And I'll tell you why we kind

[00:15:05] of use this term monolith.

[00:15:07] And it might sound like a

[00:15:08] pejorative term, but it's really

[00:15:09] not. Let me describe what we

[00:15:11] mean when we say model if we talk

[00:15:12] about these monolithic

[00:15:13] applications on the platform.

[00:15:15] It's really byproducts of

[00:15:18] the longevity of these

[00:15:19] applications. It's really by

[00:15:20] product of the fact that one

[00:15:22] of the core tenants of the

[00:15:22] platform is that you can have

[00:15:24] an application written 3040

[00:15:26] years ago that could still run

[00:15:27] today. And that's true.

[00:15:29] Right. But what we're seeing

[00:15:31] is sort of I guess the

[00:15:33] potential downside to that

[00:15:35] and the potential downside that

[00:15:37] we're seeing is that these

[00:15:38] applications are spanning

[00:15:40] again, in many cases, decades.

[00:15:42] And what that means is the

[00:15:43] original architectural design

[00:15:45] fidelity is lost over time.

[00:15:48] Why? Because you actually have

[00:15:50] different authors of this

[00:15:52] program over the years.

[00:15:54] And each of these authors has

[00:15:55] their own kind of design voice.

[00:15:57] They have their own

[00:15:57] architectural voice like you

[00:15:59] can and even in my history, I

[00:16:01] can look at kind of a system

[00:16:04] and I can sort of understand,

[00:16:05] hey, I think this was done by

[00:16:07] this is probably a combination

[00:16:08] of these four people that built

[00:16:10] this part of it because I

[00:16:11] know, since I've worked with

[00:16:13] them, I know how they write

[00:16:14] their code and how they

[00:16:15] organize their thoughts and

[00:16:16] how it manifests himself in

[00:16:18] the deliverable. Well, when

[00:16:19] you have 203040 people doing

[00:16:21] that over time, these

[00:16:23] applications just kind of

[00:16:24] grow and they expand.

[00:16:26] And again, you lose that

[00:16:27] original fidelity.

[00:16:28] And then you kind of had

[00:16:29] this layering motion of

[00:16:31] maybe a developer says, I want

[00:16:32] to develop this helper function

[00:16:34] that I need not really knowing

[00:16:35] that was already in this

[00:16:37] application from 12 years ago.

[00:16:39] They didn't know it.

[00:16:40] So kind of have repeated

[00:16:42] and sort of redesigned and

[00:16:43] redeveloped really kind of

[00:16:44] redundant code.

[00:16:46] And so that's kind of where

[00:16:46] this monolith just kind of

[00:16:48] grows over time.

[00:16:49] It still does everything it

[00:16:50] needs to do.

[00:16:51] But it's now kind of hard

[00:16:53] to understand all the moving

[00:16:54] parts, all of the module

[00:16:56] dependencies, all of the data

[00:16:57] dependencies that this

[00:16:59] application has.

[00:17:00] So this application understanding

[00:17:01] space, which is kind of part

[00:17:03] one or step one of our

[00:17:05] kind of gen AI motion in this

[00:17:07] space is to get that

[00:17:08] application understanding.

[00:17:10] And let me give you an example.

[00:17:11] Work with a client last year

[00:17:13] that a simple account

[00:17:15] balance read just to read

[00:17:16] the account balance of a

[00:17:17] particular individual spanned

[00:17:19] over 40 modules.

[00:17:21] That's four zero.

[00:17:22] So you can see the complexity

[00:17:24] that's introduced as he seems

[00:17:25] kind of grow over time.

[00:17:27] So I think that's a big one

[00:17:28] is application understanding.

[00:17:29] Second is really was I

[00:17:31] mentioned earlier code

[00:17:32] explanation, right?

[00:17:33] Code wasn't always

[00:17:34] necessarily documented.

[00:17:36] And if it was not documented

[00:17:37] well, so let's use gen AI

[00:17:39] to actually explain the code

[00:17:41] and do inline documentation

[00:17:43] of that code so it becomes

[00:17:45] more understandable as folks

[00:17:46] are looking at me.

[00:17:47] Heck, I've looked at Java

[00:17:49] code from 20 years ago that

[00:17:50] really could use some code

[00:17:51] explanation, right?

[00:17:52] I mean, so it's not to this

[00:17:53] is not a cobalt problem.

[00:17:54] This is a code problem.

[00:17:56] And then lastly, testing.

[00:17:58] Right.

[00:17:58] I mean, that's a huge challenge

[00:18:00] is the sort of the lack,

[00:18:02] I would say, of a really good

[00:18:05] automated testing framework

[00:18:06] and automated testing pipeline.

[00:18:08] And so those three

[00:18:09] application understanding

[00:18:11] explanation of code and testing

[00:18:12] I think are the most

[00:18:13] significant challenges.

[00:18:14] And frankly, we're leaning on

[00:18:16] gen AI to help with all three.

[00:18:18] So as this adoption

[00:18:20] of generative AI in application

[00:18:22] modernization continues

[00:18:23] to gain momentum

[00:18:25] as fast forward 12 months,

[00:18:26] 24 months, what changes

[00:18:28] do you foresee in the the roles

[00:18:30] and skill sets required

[00:18:32] with the IT teams?

[00:18:33] And how can some of those

[00:18:35] employees, how can they better

[00:18:36] prepare for that shift?

[00:18:38] Because if it's going to be

[00:18:39] different skills required now.

[00:18:41] Yeah, that's a great question.

[00:18:42] I've been I've heard that not

[00:18:44] often enough to be quite honest

[00:18:45] but if you think folks need

[00:18:47] to start thinking about this,

[00:18:48] I would say first of all,

[00:18:49] developers can rest assured,

[00:18:51] right?

[00:18:51] Because this is AI

[00:18:54] and our estimation of my

[00:18:55] estimation is a developer

[00:18:56] assisted motion.

[00:18:57] So you still need your strong

[00:18:59] developer skills and acumen

[00:19:01] to actually have this be

[00:19:01] successful to have the right

[00:19:02] marriage between a developer

[00:19:04] and kind of the AI system

[00:19:05] that's helping basically make

[00:19:07] their jobs easier.

[00:19:08] I will add though that many

[00:19:10] think that it will reduce

[00:19:13] the amount of skilled development

[00:19:14] professionals that need to

[00:19:16] exist in the market, right?

[00:19:17] They think that it gen AI is

[00:19:18] going to help maybe have

[00:19:19] lesser skilled people be

[00:19:21] amazing.

[00:19:22] I actually think the opposite.

[00:19:23] I think we're actually

[00:19:25] going to need even more

[00:19:27] kind of acute developer skills

[00:19:28] and a number of key areas to make

[00:19:30] this really successful.

[00:19:31] And I think in no small part,

[00:19:33] the ones that are more successful

[00:19:35] going to kind of lead the pack

[00:19:36] you're moving forward to the

[00:19:36] ones that are actually better

[00:19:37] application architects

[00:19:39] because they really are going

[00:19:40] to need to know what they want,

[00:19:42] how everything in their

[00:19:43] applications can going to come

[00:19:44] together, how it's all

[00:19:45] connected like the overall design

[00:19:47] really maximize their usage

[00:19:48] of the AI system capabilities

[00:19:51] that are offering in the

[00:19:51] market today.

[00:19:53] Because if you're just going

[00:19:54] to write use AI to generate

[00:19:56] like a function B, function C,

[00:19:57] function D for you,

[00:19:59] you're not you're missing out on

[00:20:00] the bigger picture.

[00:20:01] And so I think we need

[00:20:03] actually more, not maybe

[00:20:04] more skilled but tighter skilled

[00:20:06] and maybe more kind of

[00:20:07] nuanced skills in the area

[00:20:09] being a better application

[00:20:10] architect.

[00:20:10] I think we're also going to see

[00:20:12] developers going to need to be

[00:20:13] able to really interpret code

[00:20:15] on the fly in the moment

[00:20:17] that they didn't write

[00:20:18] themselves to ensure it's

[00:20:19] what they really want and

[00:20:20] what they really need is as

[00:20:21] they lean on AI to say

[00:20:23] help me with this task

[00:20:25] and the AI pumps it out.

[00:20:27] You can't just you should not

[00:20:29] just certainly you can

[00:20:30] but you should not just

[00:20:31] blindly say, yep, that's what I

[00:20:32] want. Let's go run with it.

[00:20:34] They need to be able to do

[00:20:35] that code interpretation on

[00:20:36] the flight and make sure it's

[00:20:37] really lining up with what

[00:20:38] they wanted and to help kind

[00:20:40] of massage that code as it

[00:20:41] goes into their framework,

[00:20:42] which again is a different skill

[00:20:44] than I'm going to I'm going

[00:20:46] to do this myself and I'm

[00:20:47] going to be responsible for

[00:20:48] every single line that goes

[00:20:48] in this and have intimate

[00:20:49] knowledge of every single line

[00:20:50] that goes in this.

[00:20:51] So I think that's different.

[00:20:52] The other is we need more

[00:20:53] of a security mindset.

[00:20:55] Right up front to identify

[00:20:56] these potential risks, right?

[00:20:58] With something in the AI

[00:20:59] generating actually security

[00:21:01] vulnerabilities, right?

[00:21:02] There are certainly tools

[00:21:03] to assist here, right?

[00:21:04] But we're going to need

[00:21:05] the developers to have more

[00:21:06] acumen, right?

[00:21:07] To leverage them as a part

[00:21:09] of their kind of day to day

[00:21:10] development cycles and

[00:21:11] development activities.

[00:21:12] This can't be something we

[00:21:13] put off until the end

[00:21:14] because you might end up with

[00:21:15] a ton of vulnerabilities

[00:21:17] that is really going to slow

[00:21:17] down your development cycle.

[00:21:19] And I mean, and then

[00:21:20] lastly, right?

[00:21:21] You mentioned it earlier,

[00:21:22] everything is changing so

[00:21:24] quickly all around everywhere,

[00:21:25] right? And this is what

[00:21:26] hybrid is all about.

[00:21:27] But it's all interconnected.

[00:21:29] And so we're going to have

[00:21:30] just better kind of

[00:21:32] skills around integration

[00:21:34] testing muscle in this space.

[00:21:35] Right? I mean, integration

[00:21:37] testing is goes obviously

[00:21:38] beyond unit, beyond functional

[00:21:39] testing. We need really strong

[00:21:41] kind of kind of muscle acumen

[00:21:43] and integration testing.

[00:21:44] That's going to be a big

[00:21:45] skill that we need to build

[00:21:47] up to make sure that as

[00:21:48] we're building these systems,

[00:21:49] these highly integrated systems

[00:21:51] that span multiple platforms

[00:21:52] and multiple services,

[00:21:53] we need really good integration

[00:21:55] testing frameworks in place to

[00:21:57] enable this to happen.

[00:21:57] And then all the automation

[00:21:59] that backs all of this.

[00:22:01] I'd love to bring to life

[00:22:02] everything that we're talking

[00:22:03] about here with maybe a case

[00:22:05] study or success story.

[00:22:07] And I appreciate you probably

[00:22:08] can't mention any names, but

[00:22:09] is any example you can share

[00:22:11] where IBM strategies and

[00:22:13] technologies in application

[00:22:15] modernization and

[00:22:16] generative AI have

[00:22:17] significantly benefited

[00:22:19] an enterprise?

[00:22:20] Is that an example you could

[00:22:21] possibly share?

[00:22:22] Yeah, there is.

[00:22:23] And I could actually be a client

[00:22:24] which is also good.

[00:22:25] It's always less fun when it has

[00:22:27] to be anonymous.

[00:22:28] For me, it's like,

[00:22:29] is he really telling the truth?

[00:22:31] But I am.

[00:22:32] But when you look at certainly

[00:22:35] the gen AI space is still

[00:22:37] fairly new.

[00:22:37] Like this just went live

[00:22:39] late October of last year.

[00:22:41] We have a number of

[00:22:43] this is where unfortunately I

[00:22:44] can't go in into specific

[00:22:45] clients. But what we'll say

[00:22:47] is the pipeline is

[00:22:48] significant and we have been

[00:22:50] helping a lot of clients

[00:22:51] worldwide with proof of concepts

[00:22:53] actually leaning on our

[00:22:53] technology to help them start

[00:22:55] this modernization process

[00:22:56] using gen AI techniques in

[00:22:58] their application estate.

[00:22:59] And it's incredibly

[00:23:01] encouraged by the results

[00:23:03] we are saying now, I will say

[00:23:05] in the application

[00:23:05] modernization overall space

[00:23:07] pre gen AI, which

[00:23:09] working with this client,

[00:23:11] we're starting to evolve into

[00:23:12] leveraging gen AI.

[00:23:13] I'll mention atruvia, right?

[00:23:14] Atruvia is probably the largest

[00:23:17] banking IT provider in Germany.

[00:23:18] And we've been working with them

[00:23:20] for a while.

[00:23:20] And what they wanted to do was

[00:23:22] future proof their platform

[00:23:24] future proof their their core

[00:23:25] banking framework.

[00:23:27] And so it was really a 100%

[00:23:30] mostly 100% cobalt

[00:23:32] framework.

[00:23:33] And their strategy was to go to

[00:23:34] Java.

[00:23:35] And so we worked with them

[00:23:37] for a while on helping them

[00:23:39] interoperate again between

[00:23:41] cobalt and Java in the same

[00:23:42] transaction at work.

[00:23:43] They needed to be dealing

[00:23:44] in that space.

[00:23:45] It couldn't be dealing with

[00:23:46] two face, two face commit

[00:23:48] compensating transactions, all

[00:23:49] that nonsense that really,

[00:23:51] really kind of complicates your

[00:23:52] entire application on

[00:23:54] everything that surrounds it.

[00:23:56] So they wanted this all running

[00:23:57] on Z a mix of cobalt and Java.

[00:24:00] I can tell you today in going

[00:24:02] through this process with them,

[00:24:03] they have Java enabled over

[00:24:05] 85% of their core banking

[00:24:07] framework.

[00:24:08] They run over 400 million

[00:24:10] transactions a day with peak

[00:24:12] throughput reaching over 12,000

[00:24:13] transactions per second,

[00:24:15] all leveraging a mixed language

[00:24:17] workload of Java, cobalt

[00:24:19] and at times assembler.

[00:24:20] Right? Again, all managed by the

[00:24:21] same transaction at work.

[00:24:23] Right? So it's if anything that

[00:24:25] this is demonstrating the value

[00:24:27] of what we can do with Jena AI

[00:24:28] now it's going to help make that

[00:24:29] transition from cobalt to Java

[00:24:31] even easier.

[00:24:32] What it really did was they

[00:24:33] said any new net business

[00:24:35] service, that's Java.

[00:24:37] Right? It's an exception to

[00:24:38] be anything but Java.

[00:24:40] Also, if they're going to go

[00:24:41] in meaningfully update an

[00:24:42] existing cobalt service at

[00:24:44] that time and at that time

[00:24:45] alone, do they then transform

[00:24:47] and migrate that one cobalt service

[00:24:49] to Java?

[00:24:50] So then over time, more and

[00:24:52] more of their cobalt assets are

[00:24:53] getting transformed and modernized

[00:24:54] to Java because again, they wanted

[00:24:56] that future proof to one of the

[00:24:57] agility which they have

[00:24:58] achieved in spades.

[00:24:59] They've done a great job here.

[00:25:01] But yes, it's a great story that

[00:25:03] we'd love to tell.

[00:25:03] I mean, when we have a client

[00:25:04] at that scale, right with 85%

[00:25:07] of their workload on the

[00:25:07] platform is mixed language

[00:25:09] with 400 million trans a day.

[00:25:10] Right? It's pretty significant.

[00:25:12] And again, this is where Jena

[00:25:14] AI is really going to accelerate

[00:25:15] that motion for all of our clients

[00:25:16] worldwide so they too can get

[00:25:18] there to achieve the success.

[00:25:20] Really, and I think that does

[00:25:21] such a great job of just

[00:25:22] shining a light on this and

[00:25:24] the difference that it can make.

[00:25:25] And if we can continue to look

[00:25:27] ahead, are there any other key

[00:25:29] trends that you might

[00:25:30] anticipate IT infrastructure

[00:25:32] and application development

[00:25:34] might be encountering over

[00:25:36] the next few months and years

[00:25:38] and ultimately anything that

[00:25:40] enterprises should be better

[00:25:41] positioning themselves to take

[00:25:43] advantage of in the near future?

[00:25:45] You probably get asked this a lot,

[00:25:46] but what excites you and

[00:25:48] what trends are you seeing?

[00:25:50] Yeah, for sure.

[00:25:51] Right. I mean, I think it's

[00:25:53] pretty evident that AI

[00:25:55] really is changing the way

[00:25:56] developers approach sort of

[00:25:58] problem solving in general.

[00:25:59] Right. I mean, it's automating

[00:26:01] more now than just the simple

[00:26:02] stuff. Right. It's automating

[00:26:03] more than meaningful stuff

[00:26:04] and it's going to help with

[00:26:05] optimizing what you're building,

[00:26:06] making recommendations along the way.

[00:26:08] Right. The impact of all of this

[00:26:11] is that I believe that

[00:26:12] the development lifecycle is

[00:26:13] going to be significantly reduced.

[00:26:15] Right. Both in dev and testing

[00:26:17] because there'll be more and more AI

[00:26:18] assisted as we're moving forward.

[00:26:20] I think the fallout

[00:26:21] follow sort of a negative word.

[00:26:23] I think with the benefit of this

[00:26:26] is going to be really a

[00:26:28] more significant focus on

[00:26:29] innovation and

[00:26:32] there's going to be a challenge

[00:26:33] keeping up now with the rate

[00:26:35] and pace of that innovation.

[00:26:37] So organizations need to be

[00:26:38] prepared and need to prepare

[00:26:40] the process by which they can

[00:26:41] deliver these new experiences

[00:26:43] quickly with confidence.

[00:26:45] Well, this might sound like

[00:26:46] traditional DevSecOps,

[00:26:47] but I would say in many

[00:26:50] enterprises, this isn't fully

[00:26:51] optimized. Again,

[00:26:53] earlier I mentioned kind of the

[00:26:54] risk some see with kind of

[00:26:56] continuous delivery and

[00:26:57] continuous updates. Right.

[00:26:58] Well, it's going to have to be

[00:27:00] optimized. Otherwise they are

[00:27:01] going to be left behind.

[00:27:02] Right. As other companies are

[00:27:03] innovating at a rate and pace

[00:27:05] and scale that we've never seen

[00:27:06] before because JNA helps them

[00:27:07] you do things so much more faster.

[00:27:10] So what are we talking about

[00:27:11] there? We're obviously talking

[00:27:12] about get your DevSecOps kind of

[00:27:14] kind of pipeline in place,

[00:27:16] get that in place, get that

[00:27:17] optimized. Right.

[00:27:18] I mentioned earlier,

[00:27:19] integration testing across the

[00:27:21] board. Again, this huge

[00:27:23] drain pace of innovation is not

[00:27:25] just on Z. It's going to be

[00:27:26] everywhere.

[00:27:27] Right. And all of these

[00:27:28] applications, not all, but so

[00:27:30] many of them are interconnected

[00:27:32] and in ways that they weren't

[00:27:33] previously now that we have so

[00:27:34] much options, so many options

[00:27:36] now to consume the right

[00:27:37] business service or the right

[00:27:38] data service from the right

[00:27:39] platform. Right.

[00:27:40] Iberts about like just best fit

[00:27:42] platform for that workload.

[00:27:43] So you're going to be consuming

[00:27:44] services from different platforms.

[00:27:46] Integration testing becomes

[00:27:48] very important in this space.

[00:27:49] So make sure you have the right

[00:27:51] kind of processes built around

[00:27:52] that. There's going to be a

[00:27:53] significant operational impact

[00:27:55] as well. Right.

[00:27:55] Needing to manage all of this

[00:27:57] at scale never before seen.

[00:27:59] Right. Making sure we have full

[00:28:00] operational insight across

[00:28:02] the spectrum, right.

[00:28:03] Across all of the platforms.

[00:28:05] No black boxes whatsoever.

[00:28:06] Like organizations really need

[00:28:08] to get prepared for that

[00:28:09] because it's coming.

[00:28:10] Right. And honestly, it's I'm

[00:28:12] really darn excited to see

[00:28:14] this shift occur and really what

[00:28:15] the future holds when we start

[00:28:17] seeing all that innovation come

[00:28:18] to life. Like I think it's going

[00:28:19] to be amazing.

[00:28:21] Wow. Incredibly cool.

[00:28:22] I've learned so much today

[00:28:23] around application modernization,

[00:28:25] how it allows enterprises to

[00:28:27] execute a successful hybrid

[00:28:29] cloud strategies or how IT

[00:28:31] leaders are recognizing this

[00:28:32] by its time have a little bit

[00:28:34] of fun with you now.

[00:28:35] I mentioned a few moments ago

[00:28:36] that you've probably heard and

[00:28:37] seen a lot of crazy things

[00:28:39] throughout your career.

[00:28:39] So I've been asking you to look

[00:28:41] back at that career now.

[00:28:42] What's the funniest or most

[00:28:44] interesting story that you are

[00:28:46] allowed to share that has happened

[00:28:47] in your career?

[00:28:48] I suspect there's a few you can't

[00:28:49] but is there anything you can?

[00:28:51] Yeah, no, certainly.

[00:28:52] And it's sort of sort of too

[00:28:54] come to mind and all try to be

[00:28:55] brief. There are mostly ones

[00:28:57] that that folks make fun of me

[00:28:59] for. Yeah.

[00:29:00] If you can't poke fun at

[00:29:01] yourself like what can you poke

[00:29:03] fun of. So we all travel.

[00:29:04] Many of us probably a lot more

[00:29:06] than we'd like to.

[00:29:06] And I had one trip that was

[00:29:08] rather significant in nature.

[00:29:09] I had to go to Bangkok to me

[00:29:10] with a CEO of a family owned

[00:29:12] bank. And it was over

[00:29:14] 24 hour kind of wall clock

[00:29:16] time for me to get there and

[00:29:19] let in Bangkok and spent the

[00:29:21] entire day like the entire day

[00:29:23] with the IBM team kind of

[00:29:24] preparing for this other

[00:29:26] worldly presentation that was

[00:29:27] going to be amazing.

[00:29:29] When back after a long day

[00:29:31] and about nine o'clock at

[00:29:32] night. No, not right exactly

[00:29:33] nine o'clock tonight. That's

[00:29:34] so acute this memory is for

[00:29:36] me. I'll call from the

[00:29:37] client team saying you know

[00:29:38] what. We really want this

[00:29:40] bank to sign this contract for

[00:29:42] us. It's coming up. We will

[00:29:44] want to get this contract.

[00:29:45] So I want to want to get ink on

[00:29:46] paper. And so we're kind of

[00:29:48] going to push your talk because

[00:29:49] we feel your talks more future

[00:29:51] capabilities around modernization

[00:29:53] and where they want to be

[00:29:54] investing in the future.

[00:29:55] And so we just kind of want

[00:29:57] to have the talk on security

[00:29:59] and we're going to kind of

[00:30:00] postpone yours a little bit

[00:30:01] is like OK no problem.

[00:30:02] What is it going to be?

[00:30:04] Well some other trip.

[00:30:06] So I said OK great.

[00:30:08] I woke up the next morning

[00:30:10] changed my flights and proceeded

[00:30:11] to fly 24 hours about

[00:30:13] 27 hours wall clock time

[00:30:15] to head back home.

[00:30:17] So I had over a 50 hour trip

[00:30:20] for absolutely nothing.

[00:30:23] I mean I did see some pretty

[00:30:24] cool monitor lizards like

[00:30:25] giant monitor lizards all over

[00:30:27] the place like just randomly

[00:30:28] walking around these parks

[00:30:29] which is really cool.

[00:30:31] But yeah, I get made fun of

[00:30:32] a lot as don't pull a Kyle

[00:30:34] right. I mean it's this thing

[00:30:36] that has become a

[00:30:37] become a thing now with a lot

[00:30:38] of my colleagues when they

[00:30:39] heard about that.

[00:30:40] And I'll just do the second

[00:30:42] one which isn't as funny

[00:30:43] but people still make fun

[00:30:44] of me for.

[00:30:45] I just this was only going

[00:30:47] about three quarters away

[00:30:47] across the country to go

[00:30:49] meet with the IBM local team

[00:30:52] this big plan for a kind

[00:30:53] of a wealth management

[00:30:54] modernization for a very

[00:30:55] large bank not a family owned

[00:30:58] and I get up in the morning

[00:30:59] shower head into the meeting

[00:31:01] and I sit in the reading

[00:31:02] ready to rock and roll

[00:31:03] in this gal on the other side

[00:31:05] of the room says excuse me

[00:31:06] are you wearing cologne

[00:31:08] said no I am not

[00:31:09] I just use the hotel shampoo

[00:31:12] she goes well I'm really

[00:31:14] allergic to cologne

[00:31:15] and that smells an awful

[00:31:17] like cologne awful like like

[00:31:18] cologne can you please

[00:31:19] leave the room

[00:31:21] and I said I just looked

[00:31:22] that was literally stunned

[00:31:24] like OK you guys

[00:31:25] brought me here specifically

[00:31:27] so I could help you

[00:31:29] and everyone was shocked

[00:31:30] like they were honestly

[00:31:31] aghast.

[00:31:32] So I left the room

[00:31:34] I sat out in the other room

[00:31:36] and then proceeded to ask

[00:31:38] Sir WebEx is there a zoom

[00:31:40] that I can kind of do this

[00:31:41] remotely and they said you

[00:31:42] know what you know what no

[00:31:43] this room isn't really equipped

[00:31:44] for that like we just get

[00:31:45] on the laptops

[00:31:47] in the day I just ended up

[00:31:48] going back to my hotel

[00:31:50] pretty much fuming

[00:31:51] the next day was a lot

[00:31:52] different by the way

[00:31:54] she was the one that wasn't

[00:31:55] allowed in the room

[00:31:57] and we had a WebEx

[00:31:58] that she got to like tap

[00:32:00] into for the next two days

[00:32:01] but I was it was really odd

[00:32:03] and when I shared this through

[00:32:04] they're like what the heck

[00:32:06] and I wasn't even wearing cologne

[00:32:07] like it was a very

[00:32:08] it was a very surreal experience

[00:32:10] and it will say she no longer

[00:32:12] works at IBM not because of this

[00:32:14] but but yeah that was

[00:32:16] you think that's something

[00:32:17] you should probably share in advance

[00:32:19] if there's people that you haven't

[00:32:20] met before or coming

[00:32:21] into a meeting with you

[00:32:23] that maybe you're definitely

[00:32:24] allergic to something

[00:32:25] that they might put on their bodies

[00:32:26] and inadvertently like hotel shampoo

[00:32:28] but anyway those are the two

[00:32:30] I had that I thought I'd share

[00:32:32] that just come to mind when you talk

[00:32:33] about what strange things have

[00:32:34] happened in your career

[00:32:36] Oh two absolutely fantastic

[00:32:38] stories do they do secret Santa

[00:32:40] over there in the US

[00:32:41] because if they did and I ask

[00:32:43] you I would get you cologne

[00:32:44] every year but without

[00:32:46] exactly right

[00:32:47] I mean honestly Neil

[00:32:48] I'd be disappointed if you didn't

[00:32:51] well it's been fantastic chatting

[00:32:53] with you today

[00:32:53] and for anyone listening wanting

[00:32:55] to talk about more of the

[00:32:57] serious things we've talked

[00:32:58] about today maybe want to find

[00:32:59] out more information

[00:33:00] connect with you or your team

[00:33:02] or just have a look around the IBM

[00:33:03] website which is huge

[00:33:05] where would you like to point

[00:33:06] everyone listening

[00:33:07] Oh my goodness so I would start

[00:33:08] with searching for what's the

[00:33:10] next code assistant for Z

[00:33:12] Yeah that I'll land you in a

[00:33:13] in right up front the lead

[00:33:15] page for everything that we

[00:33:16] really talked about today

[00:33:18] you'll get good full around

[00:33:19] understanding the whole life

[00:33:20] cycle experience

[00:33:22] and yeah you could engage with

[00:33:23] the team we can get you get

[00:33:24] engaged with with an on the

[00:33:25] ground POC work with you and

[00:33:27] your teams will bring our top

[00:33:29] Northcliffe engineering teams

[00:33:30] on site to actually help you

[00:33:32] in this journey so I think

[00:33:34] that's a great starting point

[00:33:35] to go to and then certainly

[00:33:37] find me out LinkedIn hit me up

[00:33:38] with any questions you got

[00:33:40] and I'll definitely be

[00:33:40] respond and look into connect

[00:33:42] Well we talked about so much

[00:33:44] there from why it spending

[00:33:46] should start with modernizing

[00:33:47] applications you're speaking

[00:33:49] my language there how developers

[00:33:51] can harness generate of AI for

[00:33:52] co-translation and modernization

[00:33:55] expectations for new evolving

[00:33:57] ways in which generate of AI

[00:33:58] will support enterprises on

[00:34:00] their modernization journeys

[00:34:01] I've got to say one of the

[00:34:02] things that I'll probably

[00:34:03] remember more than anything

[00:34:04] many years after this podcast

[00:34:06] is your your very amusing

[00:34:09] and story it's that

[00:34:11] probably didn't feel amusing

[00:34:12] at the time but they really

[00:34:14] were thank you so much for

[00:34:15] showing all that with me

[00:34:16] so really appreciate you talk

[00:34:18] Hey Neil appreciate your

[00:34:19] time as well my friend

[00:34:19] you take care

[00:34:20] Wow what an incredible guy

[00:34:23] and I meant what I said there

[00:34:24] five, 10, 15 years from

[00:34:26] here and I look back over

[00:34:28] what three, four, five thousand

[00:34:30] interviews those stories

[00:34:32] that Kyle shared that is the

[00:34:34] kind of stuff that I'll remember

[00:34:35] forever but on a more serious

[00:34:38] note as businesses gear up

[00:34:39] for faster innovation and

[00:34:41] more complex system integrations

[00:34:43] over to you what steps

[00:34:45] is your organization taking

[00:34:47] right now to harness the power

[00:34:49] of AI and its modernization

[00:34:51] efforts and if you are in

[00:34:53] one of those businesses that

[00:34:54] are still trying to pretend

[00:34:55] that it's not happening

[00:34:56] or banning it or shutting it down

[00:34:58] or put it on the

[00:35:00] will be nice to do one day

[00:35:01] let me know your story there as

[00:35:02] well I want to hear your thoughts

[00:35:04] I want you to join the

[00:35:05] conversation so we can explore

[00:35:07] together how these technologies

[00:35:09] are making a definitive

[00:35:11] impact on the IT landscape

[00:35:13] so email me tech blog writer

[00:35:15] outlook dot com Twitter

[00:35:16] LinkedIn Instagram just

[00:35:18] at me or see huge and

[00:35:20] until next time keep pushing

[00:35:21] the boundaries of what

[00:35:22] your applications can achieve

[00:35:25] I remember to share those stories

[00:35:26] but thank you for listening as always

[00:35:28] and until next time

[00:35:30] don't be a stranger