Unlocking Agility: Composable Solutions, Flexible Stacks & Hybrid Architectures

0:00:07.8 Dom Selvon: With a digital transformation, particularly a digital transformation to a more composable or flexible stack. The real meat of the work is the change that's actually going to happen to the business. It's a change that's gonna happen to the communication within the business. How people talk to each other to get things done is going to change. And that is something businesses don't accept, understand, believe, or are willing to take on.

0:00:40.4 Kailey Raymond: Hello and welcome to Good Data, Better Marketing. I'm your host, Kailey Raymond. Today, we dive deep into digital transformation trends. From the cutting edge of AI and composable systems to hybrid architectures, you'll learn invaluable insights that challenge conventional thinking and offer a glimpse into the future of technology. Stay tuned as Apply Digital's Dom Selvon highlights the importance of being agile and confronts some of the thornier topics of today's digital age concerning ethical and sustainability challenges.

0:01:52.3 Kailey Raymond: Today, I'm joined by Dom Selvon, Chief Technology Officer at Apply Digital. Dom is a technology leader with 25 years of experience in cloud, native and the MACH communities. At Apply Digital, he's committed to employing technology for commerce and content, helping businesses succeed. Dom currently serves as a member of the MACH Alliance Executive Board where he advocates for cloud adoption and agile methodologies. Dom, welcome to the show.

0:02:16.6 Dom Selvon: Thank you, Kailey. Really happy to be on and thank you for having me.

0:02:20.9 Kailey Raymond: I'm really excited for you to be here. In particular, we have a lot to talk about related to some of these kind of newer trends related to agile and composability and flexibility in the enterprise. And so really excited to hear your take on this as a technology leader. But kicking off today, I wanted to just learn a little bit more about your career. In your own words, tell me a little bit more to how you got to be the Global Chief Technology Officer at Apply Digital.

0:02:49.8 Dom Selvon: Gosh, you know, right place, right time is often the situation with these sorts of things, but a bit of hard work peppered in there as well. So I've kind of turned my hand at quite a few different roles in the technology space from consulting, from product development, you know, Internet gaming and casinos and things like that way back early in my career, finance, but kind of landed in this commerce space largely because it's complex enough to keep me entertained. You know, there's some thorny problems when you're dealing with transactional systems at scale, but also it's something that has a real impact on life. Everybody shops for good or for bad. Everybody shops and having a nice seamless experience. I like to think that I can help in that front. So you say 25 years, it's possibly even a bit longer than that. But I still think it's entertaining and it keeps me fascinated, the technology, how fast it changes and how interesting it is. So try to keep myself at the forefront of things like that as I go along as well.

0:03:55.3 Kailey Raymond: You're very right. The pace of change is unbelievable to watch. Especially over the past few years. The adoption of AI and kind of being able to actually see now some of the outputs of that and folks actually executing against that has been really unbelievable. We're gonna talk about AI for sure. I wanna first learn a little bit more about how some of the roles that you're playing first, you know, Apply Digital and also through the MACH Alliance, how are you helping companies build excellent customer experiences? Where are you connecting the dots in that journey? 

0:04:28.8 Dom Selvon: So, the mission for Apply and for the MACH Alliance is actually very well aligned. It's pretty close. Apply Digital, possibly the only systems integrator, consultancy, whatever you wanna call us at this particular point in time, which we'll come to in the future. We're probably the only one that does pure lack or pure composable. We don't sort of dabble in the monolithic space. We don't dabble in a sort of historical back office which is mostly single in a box software. It's all about the composability, it's all about the flexibility. It's all about realizing business vision with technology, but always coming from that business vision first. So how we actually build excellent customer experiences, at Apply, we've got five service lines, we've got business strategy, we've got commerce, and we've got platform products and a lab for innovation. But the first four is where we really mold and guide our clients. We figure out with the business strategy element, figure out what actually makes a business tick, what sort of workflows are working within the business, now who talks to who to make things happen. How do you actually get data from A to B, how do you get product from A to B and how do you get product from your supplier to your end consumer.

0:05:41.8 Dom Selvon: These sorts of things, it's better to understand them conceptually before you try to apply any sort of technology to realize it. And if you do that, then you'll come up with the right elements of your system, of your business that are going to be built. You'll come up with the right elements of your business that are going to be bought. Because like it or loath it, there's elements of your business that are commodities that everyone's doing that aren't going to move the needle. There are elements of your business that are unique. Identifying those unique ones and really doubling down and making them unique, not compromising on someone else's opinion, means that you'll really double down on that uniqueness against your competitors. So that's where we really work with our clients to identify what's commodity and what's unique as a business capability. And then we start to talk technology afterwards. There'll be nice combined suites of technology that'll realize the unique combination that makes a business a business, that makes an enterprise an enterprise.

0:06:41.0 Kailey Raymond: It makes perfect sense when you articulate it that way. But ultimately identifying that competitive advantage that brand might have above and beyond the other ones. And, you know, I imagine in many ways especially there's a lot of folks in commerce that are selling or doing kind of similar things. A lot of it kind of boils down to that brand loyalty, that great customer experience, and kind of the ways that you are actually applying those across channels and making it feel really seamless. We've harped on the amount of time you've been in this space, but you've seen quite a lot of trends come and go. I think a lot of them probably underpin that is how do you make sure that you're developing these excellent customer experiences with now this new technology. Can you talk to me about some of these trends that are underpinning digital transformation as we know it today at.

0:07:32.5 Dom Selvon: At the MACH Alliance and Apply Digital, we talk about composability, we talk about MACH, the acronym Microservices, API-first, Cloud-native, SaaS and Headless. It's kind of quite a technical acronym, but it's kind of a dogmatic view of MACH. But there's also a pragmatic view when you work in any sort of enterprise of any sort of scale and maturity as well. And now, I don't believe there'll be any new large scale digital transformations that won't have are leaning towards a hybrid architecture, an architecture that brings together some of the slower moving elements of an enterprise. Your ERP systems, your logistics maybe would sit not that arena with the much faster moving elements of the architecture, those that are pure MACH and are pure composable.

0:08:22.6 Dom Selvon: A lot of the sort of the more established software, let's call it that has really embraced the idea of an API, really embraced the idea of being able to automate and hook into the system through other systems, rather than having a manual person talking to the green screen or whatever you got at that particular time. So that API layer that's starting to appear on these older pieces of software means that they can really nicely play within a composable architecture and they can operate within an evented architecture even as well. And so that coming into play, that becoming more and more prevalent, I think we'll start to see the acceptance of a hybrid architecture being the norm going forwards. And those that are sort of stuck on it has to be MACH, has to be microservices. They'll start to get less and less and start to realize that actually the more pragmatic view, the more forward thinking view is these hybrid architectures.

0:09:20.4 Kailey Raymond: That makes a lot of sense. There's probably gotta be a bit of both going on, right? 

0:09:24.0 Dom Selvon: Absolutely.

0:09:25.0 Kailey Raymond: I imagine especially with this acceleration of the API layer that we're talking about, there's some challenges with making sure that all those systems actually talk to each other, right? 

0:09:36.5 Dom Selvon: It's a huge element of what the MACH Alliance is doing. It's one of our halo task forces at the moment is our interoperability task force. It's been going for a year or so. Now, we've got a good few very, very big brains working on it. And it's trying to articulate exactly how, give guidance and help on how the product vendors in the market alliance and in the composable space should talk to each other. What sort of SLAs these products should have and should be presented to the outside world, what sort of contracts they need to present to other systems that rely on similar sorts of data, what that domain model looks like, what error handling should be in place, what sort of protocols these systems should communicate using. And those interoperability standards are very keen to establish a set of standards. They might not be ISO, but they will be MACH standards that people can start to adhere to. And that stamp of authority, that stamp of legitimacy that would come from the MACH Alliance will make it so that customers, clients alike, and systems integrators as well are all talking in the same language. We've got architectural patterns that allow architects to talk in the same language without having to go into the really, really low level detail of what a particular pattern means.

0:10:55.9 Dom Selvon: They can just say the strangler pattern, they can just say event sourcing. And that describes the whole holistic view of a software algorithm or software way of putting together software. We're gonna do the same sort of thing with these interoperability standards so that MACH architects and our composable architects can all talk in the same language. And that will also rise to pre-composed solutions. We'll start to see systems that are pre-composed using these interoperability standards that allow for a bridge between the shop in the box and the flexibility and composability that you get with a best of breed suite of software. Now, the appeal of a shop in a box commerce platform or a content management system that you get the front end as well as the content management, the appeal is there, it's easier to implement. You get it from day one. They're often quite feature rich. These sort of pills are well understood. But you do have to adjust the opinion of someone else as a business. And that opinionated view of the world puts off a lot of complex enterprises. And so having these pre-composed solutions start to become more prevalent with accelerators, with bootstrapping kits and things like that, it gives them the best of both worlds to start the journey. So this is not the end game, this is the start part of the journey. So that we can start off and then you can customize it. It's a composable system. You can customize over time.

0:12:30.4 Kailey Raymond: That's interesting. It's like what you're talking about is really building the Lego brick approach to kind of building your architecture and being able to identify the best in breed for what you need as a business. And earlier, you said that you help businesses identify their competitive advantage. It's in service of that, right? I think that, if you're locking yourself into some of these solutions that might be walled gardens in a way. Like it's a really challenging thing as a business to get away from and make sure that you are building what makes sense for your business that's unique and your business is always changing as well. So you need to be able to flexibly change with the times.

0:13:12.8 Dom Selvon: I wanna go back to the Lego brick analogy. It's one that's used quite a lot. And for good reason, it's something that everybody knows and you can kind of visualize how the bricks fit together and so on and so forth. But the pre-composed solution is a little bit more than just the LEGO brick. If you think of it as a, you know, you remember those three in one LEGO boxes that you get, where you get, you can build one thing, then you can take it all apart and then build a second thing, take it apart and then build a third thing following the instructions. It's kind of like that, but it's more. So it's something that you can put together and rip apart and then put together but it's pre-built models. So it's not just one brick, it's many bricks coming together to make, to be a singular whole that then you can say, okay, I don't like this brick, how this brick looks, I want to take it out and put a brick that fits into the gap the same way, but gives me extra functionality in that same domain.

0:14:06.6 Kailey Raymond: That's perfect sense. Thank you for clarifying and making sure that we're articulating that analogy in the right way so that folks can kind of fully understand that. I guess one of the things that obviously immediately comes to mind when we're talking about interoperability and kind of connecting all of these systems is the need for really secure systems and ones that are privacy first. I'm wondering if you have any thoughts on where that trend is shaping over the next year or so.

0:14:34.8 Dom Selvon: I've talked about this in other areas. And the challenge with the MACH architecture, when you're talking about microservices in particular, but composable and to somewhat a lesser extent, is with more moving parts, with more components, you're increasing the surface area of attack. You're increasing the number of entry points into a system by which a malicious actor could cause damage. You're also distributing your data across more systems as well, increasing the potential of corruption or deletion or loss. And as a result, you need to be a lot more careful as to exactly how you secure that data, how you isolate that data from other systems, how you make it consistent and available as well. Obviously, they're less in the security privacy realm and important nonetheless.

0:15:32.9 Dom Selvon: So, the thing that I counsel my clients on when we're talking about this, what's called network partitioning, it's where you distribute and dissect a network of components into many more smaller pieces to achieve the same whole, but with more flexibility. As you partition that network, you need to have consistent ways of monitoring the system, consistent ways of securing the system across the holistic enterprise architecture as well, and consistent ways of recovering from failure, recovering from attacks and then consistent ways for preventing attacks and denial of services and things like that. So that consistency, you could have it as a single point, an API gateway coming in, and everything falls off the back of that. But all you've done is then just create a single point of failure as well. So you've got to really think long and hard about exactly how you want to expose your services to the outside world and how you want to talk internally and how you want that data to flow behind the scenes as well.

0:16:40.2 Kailey Raymond: Yeah, 100%. And kind of related to this is, we've already teased this out and mentioned this a little bit, but when we're talking about privacy, security, the observability, there you go. Hard word to do this morning of your data. It becomes even more and more important to have those standards in place when we're talking about a future of AI. So, obviously there's rumors right now around AGI happening sooner than we think. A lot of the AI labs have seen lots of people come and go and prominent stories coming out with some of these safety teams that might not always agree with some of what the companies themselves are doing. There's a lot here, but I'm wondering if you have any rant that you wanna go on or any thoughts that you have about AI over the next year or so.

0:17:40.2 Dom Selvon: I'm actually at a loss for words 'cause there's so many aspects of AI that I want to get into. I think as a technologist, you absolutely have to be bought into this AI thing. If you're not interested, if you're not excited by AI, then there's something wrong with you. But with that in mind, you do have to have a healthy amount of skepticism. You really got to bear in mind that this is technology, this is software behind the scenes. There is some element that's emergent out of the models that are being built at the moment that they're building themselves. We don't quite fully understand the exact reasons that the models are coming up with the answers that they're coming up with. But we know the weights, we know the construction of these neural networks and so on and so forth. So much more than we know what's going on and our own brains. But Sam Altman's hinted that AGI is next year. I think it was a sip of the tongue in a Meet interview that he did. I think he was trying to pull that back, but he's obviously got a lot of insight into what's coming out in GPT5 and so on.

0:18:40.0 Dom Selvon: We've seen the reasoning that we've encountered with GPT401, and that's quite impressive. It's really quite impressive. And some of the software that people are building with AI, with the models that are out there and what Meta, and what AWS, what Amazon anthropic and what Google and what OpenAI are doing, it's actually quite scary and quite humbling. So that skepticism that you need to have in place is around safety. You have to have a level of safety. I don't know whether it was Sam Altman or whether it was Kame or someone like that who said, "For every technical advancement you get in AI, there has to be two safety advancements." And I do think that that's important. But safety is one thing. The ethics angle is another thing that you really need to think about. There's the data that these models are trained on. It's mostly human data. It's data written by humans. And humans, as we know, are inherently biased, whatever way that goes. And the data that's been on the Internet, you know, since '95 and maybe even earlier, has largely been written by males.

0:19:53.7 Dom Selvon: Up until now, recently that started to even out but the vast corpus of data from which these LLMs are trained on is still male, and that has the male bias. A lot of the standards, a lot of the safety elements of the wide world that we live in is biased towards males as well. And as a result, our AI systems are going to be biased still. That's something that I don't know how we're gonna get out of. That's something that we need to be aware of with our systems. But that bias and that potential lack of fairness is something that really needs to be looked at. It absolutely needs to be looked at. I think we have a risk of the costs going through the roof as well. You know, there's, everything is a subscription at the moment. Everything is a subscription regardless of how much you use it. You're paying 20 bucks a month for this, 20 bucks a month there for that, and that'll rapidly pile up for no real tangible benefit. So as an individual, you've got to be absolutely on it. But as a business, where these sort of mini subscriptions will be lost in the ether, you've got to be so aware of it.

0:21:08.3 Kailey Raymond: And there's so much there. And I think that one of the things that there's this thread that you're pulling too is like, yeah, there's the cost, absolutely. There's the subscription fees. And then there's also the cost of compute on the environment. Let's be really clear is there's this kind of like sustainability element that we're talking about as it relates to AI that is going to talk about like the Paris Agreement and what's happening right now with leaders in the world talking about how we're not even close to the standards that we're put into place. It's going to be a whole different world now when we're talking about how AI and the vast amounts of power that is needed for these incredibly intelligent systems. I don't know where we're gonna find that. We're gonna have to figure out more sustainable solutions to this. If this is truly where the world is going and we're going to be run on all of these AIs.

0:22:06.0 Dom Selvon: You've nailed it, Kailey. I think the sheer power required to do these, the training element of the LLMs, it's got better, it's got less. You know, we go back to 3.5 and 2 of OpenAI and the GPTs, they were costing upwards of like multiple billions of dollars to create the models, to create the transformers, and that's much less now. That's many, many orders of magnitude less, but the compute required to service the people using it. So the training has got less, but the people using these models has got more. And these models are offline models. They're not things that happen sort of on a request by request basis. So you send your request, the model does its crunching and spits out the answer and it goes back out. And if you've ever tried to run a model locally on a machine, no matter how powerful, I've got a pretty powerful PC here trying to run a model locally and it is tediously slow. It is tediously slow. And then you think about how fast ChatGPT, for example, responds, you must realize the amount of commute that goes behind that. And unless we start thinking seriously about how we're going to either offset that or come up with more renewable ways of producing that energy, solar, wind, and geothermal, whatever that looks like, then we're gonna be in a bad place. We're already in a bad place with fossil fuels. And I think we have a duty of care for the human race as technologists to think in terms of sustainability, not just in terms of functionality.

0:23:50.7 Kailey Raymond: And it's interesting is, I do think that I've been working in tech my entire career and I do often think that a lot of folks in tech kind of think of themselves as helping the world, saving the world in some cases. And for a long time I think that morality behind some of these larger issues wasn't as in your face. And now I do think that this is kind of one of the times in which it's right there and it's something that we're going to need to confront as an industry and make sure that we are kind of setting standards for ourselves and the rest of the world associated with this because frankly, technology leaders are the ones that are going to adopt and implement AI first and other industries will follow suit. So hopefully, we can put those standards in place. All right, we're gonna shake it off. We're gonna shake off our perhaps like cautious cautionary tale, perhaps leaning pessimistic view on AI right now and we're going to make a turn and talk about some of the things we're interested in, some of the things that we're seeing that are practical applications. And so, as it relates to AI, I'm wondering how you think about its intersection with the concept of composability that we've been talking about this morning. What are the things within that intersection that are happening right now and what are the things within that intersection that make you excited? 

0:25:13.4 Dom Selvon: So one of the remits of my job is to make sure that we have the most efficient engineering workforce that we can. One of the most tedious aspects of the systems integrator job is doing the plumbing from one system to another system, making sure that this data attribute maps correctly to that data attribute. And that is something that has historically required a human because there's some nuance in saying, you know, does first name map to name over here or does first name plus last name map to name over here? There's some intelligent nuance that needs to happen. That's a very trivial example, but there are obviously a lot more and more complex examples. But this is something that we're starting to see LON's being able to do and these generative transformers able to do for us. And a lot of the co-pilots that are helping engineers code their systems. A lot of the canvas that GPT has come out with is really quite good v0. I was at a conference with Vercel just recently. They were talking about their v0 application and that's stunning. It really is stunning about what it can produce.

0:26:30.8 Dom Selvon: So these sort of helpers to do the tedious work of mapping, it'll get more and more prevalent to the point where the classic systems integrator is probably going to be a thing of the past, and the classic design agency as well I think will possibly be a thing of the past. Because you can take something from a concept to code in an automatic fashion now and you can take something from this system provides this contract, this API contract, this system provides that API contract. This is the data model that sits between the two, put them together, please. AI can do that now. So these sort of elements of mappings and integration, but also design translation, you can even do it from a video, you can do it from an image. It's incredible. Is where things are going and that's exciting certainly, but it's also quite scary from a systems integrated perspective. But I think if you don't embrace the future, then you get lost in the past and we certainly won't be doing that.

0:27:36.6 Kailey Raymond: That's really interesting. It's something that obviously we've been working on here at Twilio with our segment CDP product, which is that exact concept of making sure that that time to value is as quick as humanly possible with the integrations and the setup being some of the most challenging part often when a user is coming in and trying to map all of their events and have the right naming, all of that. And so, leveraging AI copilots within that process to almost magically map those things together is something that is saving folks a ton of time. You mentioned that like this is almost making the role of systems integrators obsolete. I don't wanna say something like such a massive word, but I'm wondering like if you have a thought on what the future of that looks like. I mean, obviously, there's a massive strategy and creativity and expertise that you lend to this as well. Like what is that role that you're playing and what's kind of that the future of that relationship moving forward? You have any thoughts on that? 

0:28:40.8 Dom Selvon: Exactly. There's a risk of over digitization in this space and I think if we just left everything to the computers, it'll be a very bland existence that we work in. I think if we don't keep that human in the loop, that human oversight, then there's gonna be a risk of the data going skew and unexpected things coming back out of the data and out of the interactions with the customers that the brands don't necessarily want. A lot of brands are very precious about their IP, about their brand language and things like that. So I think we're a long way from just handing the reins over to the computer. I think if we don't put customer needs front and center asking, is a digital solution truly beneficial for the customer? Then we've got to think long and hard about that and user research will really help there. So making sure that A/B testing is still in place, but also getting people into a room and saying, do you like this or do you like that? What does this make you feel? What does that make you feel? I think that user testing is not gonna go in.

0:29:44.4 Dom Selvon: Neither should it, 'cause it is, we are humans and we are continuously evolving. Our likes and our needs and our expectations are continuously changing and the computer can't keep up with that. The computer can look back historically but it can't react as quick as we can yet. I think maintaining simplicity is something that will also help here as well. If you have a system that is overly complex, that's going to inherently bring in friction. If we can simplify those digital offerings to make them more intuitive, more easy to navigate, I challenge an AI to come up with something that Jony Ive comes up with. And something as just as simple and as flowing as the iPhone. It's gonna be difficult for a computer to come up with that. So having that simplicity that's providing a rich digital experience, but one that's intuitive is going to be very important as well.

0:30:50.3 Kailey Raymond: I am like, yes. And I actually have in my brain right now, I have in the back and I can't quite access it now, so maybe you'll be able to help me Dom is there are all these examples in which perhaps a person has told a computer to complete a task and it's tasks that a human would be able to complete relatively simply like "Oh, fill this glass to the highest amount that you can with water or whatever." And it decides to do all of these really odd, bizarre, complex things and it gets to the solution but in a way that it just like why would you decide to do that? So like the computer brain, you're right, isn't as intuitive as the human brain quite yet and hopefully ever. I don't know, I'm skeptical.

0:32:28.4 Kailey Raymond: A lot of what we're talking about today is obviously not an easy thing to kind of keep up to date with. It's not an easy thing to ensure that you are able to adapt to the changing needs of your customers within the very rapidly changing technology environment that we're also living in. Those two things colliding. I'm wondering if you have a thought on what some of the biggest challenges are in building great customer experiences. Any insights from any of your clients or customers or any kind of trends that you're seeing in terms of what those big meaty challenges are when they're building customer journeys and experiences? 

0:33:10.5 Dom Selvon: One of the biggest challenges that I've noticed, having built quite a few of these or done quite a few of these digital transformations, is the expectation or the thinking that these digital transformations, like they were in the past when we moved from one platform to another platform, it's an inherently technical exercise that the business will, to all intents and purposes, continue to run and operate the same way as it did previously, just with faster software, with more features, with better data, but actually with a digital transformation, particularly a digital transformation to a more composable or flexible stack. The real meat of the work is the change that's actually going to happen to the business. It's a change that's gonna happen to the communication within the business. How people talk to each other to get things done is going to change. And that is something businesses don't accept, understand, believe, or are willing to take on. And as a result, there's a massive friction between the digital transformation, the software trying to realize a new way of operating, and the business which is still moving in a sort of older direction. Unless you can have senior stakeholders pushing it forward, unless you can have people truly bought into this difficult change to the operating model, then there's always going to be a risk to these sorts of deliveries. I can't stress enough how big a challenge this or adopting this change is.

0:35:00.7 Kailey Raymond: It's funny is, it probably doesn't surprise you because you talk to these kind of senior leaders all the time in your day to day. That people element of this question is perhaps the most frequent answer on this show, it just goes to show that everybody inherently gets it, but perhaps a lot of people just it's really hard. And so, it's the harder thing to actually change within your organization is the culture and the mindshare. And it's maybe slower, especially depending on your size. And so, doing that hard work behind the scenes of making sure that you are convincing hearts and minds is in fact sometimes the hardest work you're gonna do every day. It's not actually like the hands on keyboard.

0:35:48.8 Dom Selvon: There's a big sort of amber flag or red flag where people say, that's how we've always done it. This is how we've always done it. This is how this works. It ain't broke kind of thing. And we're not trying to cancel these large, complex enterprise that it's broken. We're trying to counsel these large enterprise that there's a more efficient way to do things. And I hark back to right at the beginning of our conversation where we were talking about figuring out the business capabilities first, figuring out the way people talk to each other to make things happen within the business. If you get people talking about concepts rather than technology, then they can start to realize that actually the technology is forcing us to talk in this way. But actually what we really want to do is talk in this way. And so let's figure out what actually would be the nice streamlined view of how the business can operate and then the technology bow to that will, not the other way around.

0:36:47.6 Kailey Raymond: Yeah, technology is not a panacea for your problems. You have to be able to articulate and define your use cases and your hypotheses first and then build accordingly, not the other way around. I'm wondering, you have a ton of clients that you've kind of brought through these experiences and kind of building out exceptional customer engagement. And I'm wondering if you have any examples that you wanted to highlight that were just these kind of delightful programs or moments that the technology helped aid in building. Anything that you wanted to share? 

0:37:23.6 Dom Selvon: There's a few. I mean, we have this lab that we love to play these sorts of things, whether that's with AI, whether that's with Apple Vision or whatever. But a couple of the examples that really spring to mind in terms of customer experience really stem around discovery and conversational commerce, how you actually get to what you want to do. If you don't fully know what you're looking for, Amazon search is a hot mess. It's a nightmare to actually find anything in there. But what we've done with a company called Arc'teryx is a jacket finder where you're talking about do you want a hood? Are you going out into, is it going to be really cold or is this, you know, needs to be windproof or you know, these sorts of things like you kind of think, is that a faceted navigation, but it's a natural language conversation you're having with the system to eventually get to a recommended jacket. Similar to what you have in a shop with a very knowledgeable sales. And translating that is something that I've been wanting to do for a while. I've always got, I've called it the savvy road problem. Trying to get that little black book of my best clients saying this person likes this, this and this. This person likes this, this and this and be able to refer back to that true hyper personalization.

0:38:39.7 Dom Selvon: But being able to then articulate and take people through a process to get to the perfect garment or the perfect compute or whatever is something that Arc'teryx have done extremely well. But I think we'll start to see in a lot more, a lot more systems as well. So another example would have been what we did with Kraft Heinz and their Lunchables product where Lunchables, you get the cheese, you get the ham, you get the biscuits and things like that. And we used AI to create monsters and create machines and aliens and things like that just based off the elements of the Lunchable and it was really quite cool. It's useless, absolutely useless. Has no practical application, but it's great fun. And so whilst it's not a good example of a great customer experience, it's certainly from a brand elevation perspective, it's huge.

0:39:32.0 Kailey Raymond: It is though, right? I mean, it's like only 5% of folks and maybe this is a B2B stat, but like are in the market to purchase your products at any given time. And so making sure that you're investing in that brand and doing these like maybe oddball, unique, whatever kind of activations that are delightful are just exactly what folks need. And I saw like the smallest example of this. You know, obviously there was an election that happened in the states and certain folks feel a certain way about it and Slack in their app update just took a moment to say like, hey, no big updates. Just take a breath, walk out in the sun. They didn't say anything and they literally just said take care of yourself in their app update. And it's just kind of this delightful, beautiful little brand thing that they did that I think holds a lot of power. So it's stuff like that that I think kind of also brings back the humanity and this technology and the fun, the playfulness.

0:40:31.0 Dom Selvon: Yeah. We will get massively sidetracked if we get into politics, so I think we probably should step, side step that one very firmly.

[laughter]

0:40:40.3 Kailey Raymond: I promise we won't. The last couple questions that I wanna ask you Dom, today are the whole conversation that we've been having are really framed of course, around some of these key elements of MACH, right? So is it flexible? Is it composable? API-first, cloud-first, that kind of thing. And so I'm wondering, there's a lot of things that come and go in the tech world, in any world, in any industry. There's a lot of kind of fleeting trends. I'm wondering why are these not? Why is this concept around kind of composability, interoperability? Why is this not something that is going to leave our psyche in the next two years? Why is it here to stay? 

0:41:24.9 Dom Selvon: So why are these here to stay? The words that you used in your statement there were flexibility, composability. And we talked about interoperability before. All words talking about things working nicely, playing nicely with each other. And I think a system that has reduced friction, a system that has a smoothness to it, is always going to be beneficial to one that has friction or one that has limitations to it as well. So as a result, I think whilst it is more complex, the composable architecture is arguably more complex. There's more moving parts, there's more things talking to each other, there's more data flowing over. The flexibility that it's in part will mean that it'll stick around for a while. Colleague of mine, Neil Trickett has always said when you're doing these sorts of transformations, you're not replatforming, you're un-platforming, you're taking the systems and moving yourself away from that three, five year cycle of moving from bigger, better, bigger, better, bigger, better to little step changes that are enhancing small parts of the business and iteratively building upon it.

0:42:41.0 Dom Selvon: And that flexibility, you wouldn't be able to go back from having that flexibility to not having that flexibility. So you have the composable aspect of it. I think what you gain from that is not having to count out to someone else's opinion. I think being able to apply your own opinion, you know, what has made your business unique and successful within the particular sphere in which you operate, being able to be in charge of that, be in charge of your own destiny on that front is incredibly powerful. And I think large enterprises and medium enterprises are starting to recognize that and are starting to be more discerning about the software that they use.

0:43:26.9 Kailey Raymond: That's great. That makes perfect sense to me. You're right. Is like it perhaps is more complex to use your words, but the gains that you have for the longer term, I do think outweigh some of perhaps that shorter term cost of perhaps making sure that all these kind of things work together is you'll be able to innovate probably faster in the future if you do it right today. Last question for you today, Dom. If you have any steps, recommendations for somebody that is looking to uplevel their digital strategies, what would they be? 

0:44:06.7 Dom Selvon: I've kind of hopped on about it a little bit because I believe it so passionately. But you've got to think about the business first. You can't take technology and say this sort of seems like what we do, but not quite. You've got to think about the business first. What makes you unique? What elements of your business are commoditized? Let's lean towards buy for those commodity plays and build for those USPS so that we can be in charge of what our opinion is and how we want to present ourselves to the outside world. And I think we talked about it previously, but you've got to remember this is an operational change as well. This is not a technology change alone. There is technology involved, but technology is the easy part. Realistically, the operational change is where you'll have the most pushback. Like it all over. You will have pushback on it, but you also get the most efficiencies by embracing that operational change as well.

0:45:01.0 Kailey Raymond: So it's almost like it's the business and it's the people first and the technology is the catalyst for that change.

0:45:10.0 Dom Selvon: People and processes. People and processes. It's not going away. It's not going away. It's still people and processes.

0:45:15.6 Kailey Raymond: The unsexy stuff is always the stuff that gets buried for sure. Dom, this has been great. I learned a ton. We got to go on some fun rants too, which is exciting. So thank you for being here. Thanks for the time.

0:45:28.3 Dom Selvon: Appreciate it. Thank you so much for having me on. Really enjoyed the conversation.