Composable without Compromise w/ Dom Selvon

that's the wrong button [Music] [Music] [Music] all right we are live Dom welcome how are you I'm just rocking out to the intro music exactly that's exactly the reason why I put that in because it's so much fun to have that in there and then get started and then talk about tech stuff it's kind of a funny curve right there man um so um please say your name the way it's supposed to be said because I might not be able to say it the right way I think with yours I should be okay but maybe go for that been doing that for a good 10 years now so well yeah a bit longer than it exactly and so um recently on LinkedIn this whole composable stuff has been exploding of course between you and me we've been doing this Mac Alliance thing for a while so for us it's quite normal but I'm seeing more and more companies think about like hey how to connect headless stuff and how does all that work and then you see on the one end the more idealistic people that say if you just do good code it connects really well it's hyper component that is it's really cool and on the other hand people that say no but that's glue code and that that really hard to deal with and it's it's all over the place and so new product categories are you know kicking up and happening and so what have you seen lately that you think is interesting or and that can be positive or negative I I think it's true it's quite a new space and composability orchestration choreography digital experience all these sort of terms to become more and more entering more and more the languages not necessarily being more understood what we can do later but certainly entering into the end the language and into the market yeah what they actually mean is it's quite difficult there are early adopters I think you know those that are in the Mac Alliance of 70 80 members now and it's getting getting a real grants for their support but uh you know those that are in that Alliance really do kind of live and breathe the composable the composable way the Mac way but they are subtly different I think as well you know these these these um these vendors that have given out composed on the software a different to the the clients that are adopting a Mac way or a composable way as well you know we we did see with covid um one of the uh the sort of consequences of covert was those that had an agile business model an agile operating model they were able to react to the circumstances those that were more rigid weren't able to react to the circumstances now a lot of people put that down to having a split front end and back end or the ability to to modify your your customer experience and leave the the slower back end systems intact to sort of thing with it with yeah the many hands behind it the swan analogy you know the beautifully graceful on the top of their water and then exactly Maniac below the water but yeah I I think that was largely true you know those companies that had adopted a level of composability not composability because maybe we see it nowadays but a level of composability really thrived during 2020 2021 and it's starting to become more formalized now so he was little little understood back then very very new back then you know started in 2020 2021 yeah I was there at the beginning yeah me too I was at the company that founded it and I was I had some very rigid opinions on it that changed fast once it actually starts yes I I had a good time back in 2012 2012 when I first joined e2x and I was working with a client in um in Scandinavia they were proposing to it was related Oracle Commerce monolithic platform set the scene for a good good many years and and you know there's still a lot of Enterprise clients who are still on atg but you know back then it was it was relatively modern it was It was kind of the de facto standard for Enterprise Commerce and they were proposing taking the head off the Commerce platform and having a bespoke presentation layer and using atg just for functionality and just for data and I fought it vehemently because I just couldn't understand why you'd want to complicate things I couldn't understand why you'd want to make life more difficult for yourself you know more moving Parts more things to manage in a deployment um cycle yeah a more operating changes you know operational model will change things I don't know and I fought it I thought it I fought it until I saw it in action and until I saw the reactive reactive nature of that front end without having to impact that back end then that's the last thing actually there's something in this there's something in this yeah that was 10 12 years ago now but imagine that I had exactly that same thing with atg with a huge Global brand with like 60 markets on atg with a few other things on top and this was 2019 and I had the same thing it's like why would you change it now those Java templates or whatever you call it they have been working turns out they hadn't been because if you had to do one CSS change like a little styling thing you had to redeploy this took you know 40 hours it was a serious deployment and then at one point the CIO I think of this company said I want to deploy every week I want to be agile I want to show my shareholders how fast we can be and then the company we worked at was literally just dealing with the deploys every week not building anything it was just impossible and so we separated it out made a fancy new kind of front end and the thing accelerated that Innovation like ridiculously so there there is something to say for it but it was hard I hired 20 people took us a year right but there you go now it's amazing so there's stories out there that people have done this transition to yeah to composable in in very fast time scales you know it's it's kind of possible but you don't get much richness exactly an MVP it can be done quite quickly but then if if uh something like and this is where it gets harder I think when we talk about composable and and putting things together when you throw scale at it yes and a lot of different types of compositions start to happen between regions for example then suddenly it gets really complicated and then the developers even if they're like the most amazing ever they're going to just be dealing with these Integrations more than actually doing the Innovative storytelling that you want them to be doing what they're really good at or making things faster or cheaper to run or something if you're only dealing with the complications and so that's where it is I'm not sure if it came from that but this whole dxc thing so digital experience composition somehow grew out of that and so how have you seen that DXE thing come up is it just uniform shouting it or is there also other reasons or other you know venues where this came out of I think it was a reaction to the the sort of concept of a digital experience platform and this this opinion of how you comprise an experience or how you present an experience to the outside world you know do you take the opinion of a vendor or do you take your own opinion that's that's aligned yeah your van your brand now the vendors have been doing this for a good long time so their opinion has moved no doubt about that whatsoever but the opinion is probably a little bit generalized that's possibly a little bit you know and I say this with the the fondest nature it's probably the most common denominator as well yeah you know so it's it's it's it's it's made at a point where it covers off a good spectrum of functionality to a good enough that sounds like 80 of the market which is kind of fine that's where these things shine right exactly you know the Pareto the Pareto Principle also says 80 20. you know you know you get 8 coverage and you and you'll you'll be okay that 20 is going to take 80 more of the time so it's difficult to cover off those things but realistically as a brand you want to differentiate it's an open highly competitive review you want to make sure that what you're doing is is unique and it's something you can shout about and say you know look this is why you should come and purchase remotes or this is why you should come and read our media you know yeah you should be a differentiator and to be able to differentiate you want to be able to do the simple stuff that commoditize stuff good enough but you want to be able to do bits that make you unique extremely well you want to be able to focus your time your resources your effort your people your talent on those little areas that make you unique now with a um glue let's use your glued monolith or a modular monolith or however you want to call this whatever you want to call it the same thing you don't really have the the chance to focus on one particular area because you're you know constrained by the entire monolith again you know the the the village of modules that you're working with you're constrained by that whereas with a composed architecture that's being orchestrated properly you're able to say okay these bits I'll let them work as they're meant to out of the blocks I don't want to touch them you know let's use the common practice of those areas but for these three or these four little areas these four new modules I want to really dedicate my time to I want to give them the extra um processing power I want to dedicate my engineering teams to to enhancing the functionality on them I want to really um do lots of market research and put the user experience on each of these and bring in the data group multiple different sources and you know to be able to do that you need to be able to yeah because interestingly if you think about it that dxp thing that we used to have in these bigger systems that we now call Legacy they're still doing pretty well to be honest we don't have to call them that they're fine the thing is that the fact that the content editor could do so many things or apparently could if developers didn't mangle it in such a way that it suddenly is complicated that is a really lofty goal to have to actually have these content editors be very happy but the developers as well because happy teams just build better things yeah and of course when you compose something and you don't have a central way to somehow content edit something without having to understand 15 systems that is that's I think that's the only thing that was lacking because me as a developer I love this whole headless thing we connect everything together we make data members everybody's happy well we think they are because actually content editors or maybe who people who are in marketing or whatever they have a very high threshold for pain so the advantages deal with it until it doesn't work anymore and so it took a while for for also like you know bigger companies to understand that this approach it's super cool but me and it's cheaper but maybe it's also quite complicated for for everybody in the team so quick question here so let me just put that here as a little Banner so when going you know working at an agency being the CTO you have probably done some like one of those transitional projects from monolux monoliths to you know composable architectures so do you have like any sort of unexpected outcomes that you saw from that and like for example um one of our other guests said that they would expect that a composable architecture has 10 times faster performance on the front end which was completely not true it just became more complicated and they had to hire more Specialists to do it so that was a surprising outcome for them so what have you heard or what have you you know seen there I think is it surprising you know in hindsight on the 2020 hindsight it's pretty obvious but data actually isn't where you expect it to be if you think about the monolith the monolith has often just a single repository of data so every single system has access to that data store you know it's internally consistent and it's transactional it's um it's it's you know horizontally scaled it's it's you know right there in a composable Arena you've got to think long and hard about where the data is at any given point in time what service needs access to what data you want to be careful about over fetching the under fetching oh yeah and make sure that your data is consistent and it's available you know in the internet age people absolutely want availability over consistency they'd rather have the data there in front of them and it'd be eventually consistent they'd have to wait it's many many studies and things that's a really interesting one so even if it's dirty data there was somewhat copied from somewhere else as long as it's available that's what they want they want available data more than consistent data and as long as that data is consistent eventually as well there are situations where they'll want it highly consistent you know during a transaction yeah exactly during the checkout they're more tolerant for Less availability compared to consistency but during you know content consumption and things like that it's they'd rather availability and so what is your opinion on this idea where all these different best of breed or best of need systems whatever you want to call them where they kind of host your domain data of that specific thing that they're dealing with which means your data is actually not Central but it's completely decentralized and the tools that are smart about doing let's say digital Asset Management they deal with all the media and then your you know your Bim system has all your product data and then elsewhere you connect that image to whatever that Bim system has that is kind of decentralized so what is your opinion there has that ever worked for you or maybe there hasn't been a system to organize that well yeah as soon as you as soon as you decouple things and you have integrity problems the Integrity problems very very hard to uh to overcome and that that's not going to go away with the internet you know the um the the decoupling of systems is getting further away you know the Advent of rest meant that you have to go over a protocol to go and get information you can't just go to a database to go and get it exactly that's totally different now true ogbc it's much much faster to go over um an odbc Bridge even though that was slow compared to direct data access or you mean like the files that you directly get from the from the disk right yeah exactly yeah yeah so you know it's adding extra layers and extra layers you know um to access this data and that's that's um dislocation of information means it's it's just slower to get to it and you and transactional integrity across those boundaries is even part of there's a lot of patterns and that you can leverage to to maintain consistency there's there's things like event sourcing um transactional outbox cqrs various other things that allow you to maintain the Integrity of your denter in a read context of Integrity of data in a right context and the Integrity of your data across multiple subscribers of information as well maybe going into a bit too much technical detail here but it's important that's all good I'm for it you um it's one of the other things that maybe people don't quite grasp or or realize when going to go into this sort of um Paradigm is that the the knowledge required to achieve success in this sort of thing is actually very very difficult you've got to understand the data intimately you've got to understand the counter play of all the different moving Parts within your ecosystem well because some you'll have control over some of your words the very definition of SAS is it someone else's problem oh yeah and that is a perfect Bridge because my next question is based on your opinion with this like if there's any sort of advice you have to give to maybe like an agency exact like you are or maybe your CEO or something or to other SAS company XX that think like this okay I have to not take care of this certain piece of data which is like vital to this company's you know success like what would you advise them in the context of this whole composable thing you've got to be you gotta you gotta pay attention to interoperability interoperability is going to be one of the big keywords of 23 in my opinion and Ambience and the all these vendors who are playing in this space and obviously want to have a larger slice of the pie but that that pie is still a pie that needs to to work yeah so working playing nicely with others is going to be massively important if you can't present your data in a manner that's you know tolerant of other systems and then you're going to be rapidly moved out of the uh out of the equation yeah people are not gonna like you when that happens because you also kind of see sometimes if you have a product that works in a certain way it's more sticky to the customer yeah right specific way so if you have to do it this way and otherwise you're out which is actually pretty smart business wise but maybe there's also another type of business-wise smartness as in being so agile that you can remove that thing immediately so you have more time to do other things but then it's harder to make deals with Partners again so it's really finding that balance I guess yeah I think in the composable space if it depends on what sort of SAS business you want to be if you want to be a SAS business that owns a lot of the um the real estate within a customer's Enterprise then being agnostic being and generic enough to be you know truly composable so is is not that easy but if you aren't then you're either very sticky and then you get you get locked in as a clients and clients don't really want that these days they don't necessarily realize that once a system is deployed into their estate it's relatively sticky at that point anyway because they've invested time in it they've invested everything to it and so and so forth however if the the SAS vendor's opinion is so rigid that you can't actually customize it and bend it to the the client's unique brand vision then that's not going to work it's that's the hardest bit of this now that I work at a startup with a product thinking of these things like the one thing I've learned for example is when you say to people when you use this tool you have the freedom of choice you can do anything you want there's no conversion but when you get them into some sort of the cold feeling of this thing they buy in and then they have freedom within that choice then they're happy but that's a learning you have to have like me as a developer relations person like how do you say stuff about a product for people to interact with it so just saying it's fully agnostic it's that just just doesn't work yeah it's not I mean I think you triggered something in my mind there that the the the SAS companies the companies that are meant to work in in this ecosystem have to work in a developer ecosystem but they also have to work in a business ecosystem an operating model as well and so if it's if it's purely purely developer experience LED then that's great in some respects but businesses aren't run by developers people who have a vision and that entrepreneurial vision for building their business now so if you're if your SAS products is not usable it's not um sympathetic to the content editors then you're going to have trouble but it also needs to be a great developer experience as well because still I'm you know best will in the world no code no code um listen to something that um our mutual friend Adam Stewart was saying today about Halo oh yeah that's a whole new one again yeah another another acronym they've got to learn as well there you go all these all these things they're young they're mature the developer is still relevant in these in these spaces so the developer experience has to be as good as the content editor experience as well well the one thing I know of having lab product teams for of developers and agencies is if they don't like a tool they'll complain until they don't have the tool anymore that is the one thing I know and the more senior they get the more they complain because the more they see in the overview of why it's not working and they have to go to your level to no longer complain but it takes a long time to get there so if you have something that doesn't make them complain at least about that part they can focus that energy elsewhere so it is still a vital thing so um when you look at the space right now would there be like a Magic Wand Action for you to change something and if there is what would it be yeah I think an easy way to describe the concepts oh yeah or maybe like a blueprint to just just use this it's cool yes this is in every single situation this will work for you and that'll be that'll be beautiful I think exactly I think no I mentioned it earlier interoperability if we could have a standard protocol for all the vendors in in various different verticals that would be wonderful that would be wonderful yeah and because what we're what we're seeing now at least in my opinion is um it becomes harder to choose so there's more knowledge needed at the client side rather than the agency side to actually take ownership of these choices and then you see quite often that people are actually okay I have this job for another year I'm gonna complicate myself do I now have to sell this to my tech boss or do we just remain here and then get a better quarter next quarter because I didn't have to invest like these interesting things start to happen right so if we had just a blueprint of somehow to like make this decision this would be cool yeah I think the um one of the the shocking things that that businesses that I'm from the si side of things early early in the engagement early in this yeah transition process people think this is a tech problem yeah this transition that you uh you try and play is purely technical it's not purely to everything it's actually a business problem because you're talking about your communication Pathways you're changing that operating model internally uh so when when we work with clients we often have a process we um have a workflow to try and get them to understand that this is an operating change that they need to understand how they work now with the systems that they have currently yeah and work out what their business work out what their current software does in those components and in those business capabilities and then look at what's available to you know if maybe they've chosen the packages to do to do various aspects of their new world whether it's Commerce whether it's content research whether it's I know promotions or whatever yeah and then I think you know on their current capabilities and these proposed components that they want to have in place uh is there a one-to-one I think there never is no never is yes so those gaps you need to identify what those gaps are and those gaps could be plumbers by another SAS vendor yes that's true if that's the case then brilliant more composable if not then you need to build out your architecture to be able to plumb those gaps until it gets commoditized exactly and for that you need agencies or sis to do that right because that's where the knowledge lives like at sis it's generally the knowledge is like a few years ahead of what you actually have to do so you can advise yes um yeah and that's one of the things that we recognize relatively early in this composable um movement I suppose is that um when we were with atg it was a vendor-led environment oh yeah customers were choosing their platform because the platform did everything and then we were being introduced by the vendors now it's not a vendor-led environment it's a market lid environments so we as systems integrators need to get in there early and talk through the options through the Alternatives through the pitfalls and through the challenges and through the opportunities of going in this in this direction and that that's that's that's great it's it's it's great watching companies go through this journey and and have those those Epiphany moments and seeing how this can you know really enhanced and pushed the boundaries of what they can do and then taking it through that as well it's just it's just it's just beautiful I love how you actually called these things out because I've been talking about that exact thing but I called it platform first or Value First agencies or sis which is probably the same thing and so I've worked at companies that were there in their origin either a platform first or a value first company have you ever seen um Si's move from the one to the other or do both is that actually possible ask the question again yeah so can they be both like Fender let like still doing adobe's inside course and optimize this but also doing composable from in the same company a DNA or not I think you can but yeah but the company has to be big enough to be able to do that yeah okay yeah yeah enough people basically it's enough people I think there is [Music] um yeah I think I think they are separate enough to be able to do that I mean we we are dedicated every single bit that we do is composable software but we come from you know our and our Genesis Foundation is that monolithic world so that you know it's it's made us everything that we are exactly you go with the times I guess exactly exactly yeah and so when you think about all these modern composable stuff I think we might have touched upon this a little but have you seen any sort of bottlenecks of customers that went with this whole approach followed your advice what bottlenecks did you see like were there like process bottlenecks or maybe just deck or knowledge knowledge knowledge is the um is the biggest the biggest challenge it's it's changing the procurement is also a big challenge oh yeah yeah yeah yeah yeah that's that's a stream on its own I think we need a startup that does procurement how about that yes I'll leave that back let's let's pick that up later yeah that's that's the whole kind of words a direction maybe we don't want to go but I think um knowledge is a huge barrier bottleneck whatever you want to call it it's it's something that's and it's getting better and but being able to articulate the complexities of what's trying to be achieved with composability is actually very difficult it's not it's not it's not simple uh complex I spent an inordinate amount of my day trying to think up analogies oh yeah there's definitely Concepts that we have that resonates and it's it's really hard I mean my latest um one that I'm trying to think of to to articulate the sort of layered approach is is language sort of language you have vocabulary which is your API you have a grammar which is the orchestration or composition engine and then you have the story which is the experience and then each of those need to be comprised to make the story you need to be able to write the apis in a grammar that makes sense there's a little bit of a an early analogy and yeah you have to work to get it out but it makes sense it's it's it's getting there and I think it's starting to make sense to me and the the other one that I was I use as orchestration naturally leads to a conductor orchestrating instruments and you can move one instrument for another instrument as well so um imparting knowledge in terms of analogies is is one of the a good way of getting that across in terms of the and the architectural Concepts that we're trying to to apply I mean I I spoke I started off a number of patterns earlier and that will mean nothing to people who aren't who don't live and breathe this you know exactly even exactly and we tend to forget sometimes that this is our daily life it's like we live and breathe all this data stuff nobody else does yeah we're like the odd ones out to be honest um however I did see recently that lots of people that used to be on relatively high levels at ssis are actually moving to these product companies now or moving to clients where they are now building internal teams and so it has been getting easier I've had meetings with customers that didn't know about the the Legacy monoliths they actually like the story We Tell oh yeah you can go from this old one to the new stuff and they're like which all what what are you talking about and they're actually having issues with how they connected their new stuff right so it's slowly but surely starting to change now which is very interesting I like that I do like that it can be and the connectivity is also something that's really challenging in this in this space as well you know integration is it's never been easy it's never been fun frankly um it's it's not something that that's that's sellable as a um at all until recently I think recently we've started to see systems that that are effectively integration the fact that they are being able to be sold as as products with a UI and with a low code no code approach and with a set of patterns you know have opinionated or an opinionated they they may be is a great indicator of the maturity of this arena in which we play yeah and and also it simplifies all the other players like they don't have to now do all this integration stuff really or like simple ones because of course you need some of it but you don't really need to have cms's that fully integrate with a whole Commerce system this is going to roll up and so this this um I think you wrote an article this morning or yesterday on LinkedIn about um a lot of the um the players in this space are going to roll up into smaller and smaller and and more monopolized um companies I'd like to think that won't happen but you know it probably will yeah look at npm package manager for node.js like that's a very nerdy thing I'm sorry for the people are not as nerdy as we are but you actually have like npn packages that Jack even or odd numbers like like these crazy things and it's it's so fragmented let's see if that happens yeah well it might not be it might not be and I I don't know I'd like to think the the space will mean will maintain its vibrancy and its agility and I think it would it would lose something if it goes to a a more monopolized monolithic monopolized um landscape but I think if it does one of the benefits that will come out of that will be one of the role-ups will be a protocol but every vendor who wants to play in this space has to adhere to that that protocol and if that does happen you know I I would love that to be open oh man I would love to have like I've had these thoughts as well I won't pitch them too much because this is not the place or or the time but it would be nice to have some sort of like a file format somehow that just says this is your composition of all these tools and it's open and then every agency or Si can just make well this is the file put it somewhere in some system and suddenly your Integrations everything is there yes right and there's some there are some companies working on this but it's so complicated right now I don't think the tech is ready and there's so many different ways to skip that you can't you can't cater for all the different permutations and you also the the other challenge that you have is is it's limiting as well I mean we've talked about it earlier that you don't want to have to adhere to the lowest common denominator for the bits that make you as a brand unique that you don't care about yeah most common denominator brilliant someone else's problem don't care but the best that you do care about you absolutely want to be able to break it and uplifting exactly and that that it's not always super easy so we spoke about all this new stuff right but imagine think back in your long history what old cool feature of whatever technology it may be even in atg that you've seen would you like to bring back I I don't know page builders I think I like I liked the fact that um that page Builders were getting richer and richer that the um the screen was back in the control of the business yeah I think with the um the decomposition that we've gone through the the componentization the imposibilization and of Mark and what have you has meant that that screen ownership has has gone back into the hands of the developer exactly and you know being a developer by trade you know I like that um but in terms of flexibility growth of you know being able to focus on other things we absolutely need that screen to go back into the hands of the marketers the hands of The Visionaries the people who you know think of great new ways of merchandising and having that absolutely needs to go back into their hands except there are there are companies that are working on that are doing that very well um but it's still it's I still feel it's it's it's got a bit to catch up to where it used to be so you just without even me for coercing you did marketing for a release we did today that was not meant to happen like this but this is very cool because we actually just released something we call facial canvas which is basically it's a page builder but not like super detailed it's not like hey I'm changing this padding and merging and the color of this button but it's more I have a bunch of components that that are that match my design system and I attach sources from old places whatever API I have I attach those to those components and then you can kind of drag and drop and see what you're doing visually and I think it's not it's not fully a page builder but it's more like contextual editing so a Content editor doesn't need to know it's algolia dayquery it's just a search query because you could also query Bigcommerce or whatever and so it's it's kind of a new approach to what these page bills could be but I'm not going to talk about this too much because this is not that stream but it's funny you came up with that because I think it's a very interesting concept that sounds too coincidental to uh to not have been staged but I promise I promise this was not staged but it was perfect I've had a good conversation with some of my Architects yesterday about what a BFF should be um exactly should it know about the the um where this data comes from no it shouldn't the BFF should not care at all whether this data comes from algolia or Constructor or Commerce tools yeah from Temple or context data I'm not going to name everyone I haven't named there's so many to name Brandon I apologize but it doesn't matter because actually my BFF is an experienced buffer yeah and that BFF should be aligned to the experience for which it is presenting for which is why I kind of like graphql as well as a kind of uber BFF um but again with graphql you kind of lose the the specificity yeah I don't know if graphql is going to make it to me it feels too opinionated almost and the namings that they are using it's pretty like it graphql when you make these sorry it's going to get a bit technical again but that's what you get with guests like these uh if you talk about graphql you kind of have an amazing contract between teams if you work in a company that internally uses graphql and they all have like little products I think it's amazing it's super solid you read exactly you can see what's happening it's almost like Swagger 2.0 for internal teams but the moment you make it available to the pro to you know the public of developers I feel like it's almost limiting and opinionated and like just try to do pagination with graphql just that I wish you good luck it's very complicated and then trying to make graphql endpoints holy moly so I'm not sure it's it's a really amazing idea but it's also really hard somehow to deal with because I've used it at length I've loved it and I've hated it I've I've thought it the and the solution to all all arrows and I've also thought it's being over engineered a piece of software that puts the business logic into the presentations here again all the composition logic into the presentation too again and whilst there are benefits in there and you know when I was talking about the BFF just then but it's specifically dedicated to an experience you kind of then constraining your experience to what the BFF can provide whereas your graphql gives you a bit more flexibility in that to to choose if pick and choose whatever you want it you know if if you have you know absolute flexibility you know as as Uncle Ben says with great power cons great responsibility and if you don't you have to be mature to be able to deal with it or experience yeah yeah what I kind of like is probably a mix between these systems like having the feeling of this BFF thing where you have proper um control of all your components and doing cool stuff with that um um I think it's really cool yeah um I'm just copying your your chat quickly sorry audience you are not supposed to know this um so basically that is cool but then also that limits a little bit and so but it's also complicated to to have to query everything in your front end right if your BFF says just this is the thing queried elsewhere that also kind of sucks because if you want to do something in runtime you have 15 components that all need to query something else it's too slow yeah and so you need to have a system that which says you know what I'll cache that stuff on the edge for you if you choose that and then another system says well maybe this stuff is cached but then there's an ID to some CMS entry that I want to query myself let me do that in microservice somewhere so that's exactly what I was saying you you go into some sort of gospel people buy into it and then you give the freedom and I think that's where it wins and I think you you've touched on a couple of things there you know being able to have that data at the edge so the the speed is much better and so forth you need to be able to materialize that information from a slower systems Downstream as well and how you materialize that information is currently an engineering challenge so it could be it could be actual materialization so the data is replicated yeah risk of going stale risk of being inconsistent that is again that you need surgical cash perch and if you say catch burst to any developer they'll cringe you just twitch the the two hardest things in engineering is is caching and naming things so with that in mind let me just put um a comment from a certain CTO that has a lot of experience with doing that and then also he says biggest issue with graphql for page experience data is the inability to handle trees or arbitrary lengths which is what base data essentially is great for a lot of scenarios but not for the web experience that actually I didn't even think about that the fact that it's so variable interesting hmm it is it is it is I think um I've argued for graphql being used Downstream in in the um the back end the mid and the back of this systems and I've argued against it being used back there because the the argument against it being used further down on stack is you should know what you need and what you don't need down there it should be well sometimes yes sometimes no that's the hard bit particularly in an adventure yeah yeah sometimes yeah sometimes when an event driven architecture you kind of you shouldn't really need an interface you should have ports and adapters that respond to external influences that that's the ideal but that's never the case you always you always have static elements in your system you always have Dynamics in your system it's almost dirtied up somehow it's never the ideal thing that we would love it to be all right I think exactly and like I think we can go down this rabbit hole for very long but we're not gonna I'm gonna give you a fun question so how would you explain composable architectures to a 15 year old and I chose not like your grandma or something I chose this because these are digitally native people yeah like they know better to me than me how to use Snapchat because it has all these swipe things that are not native to me but they are native to them which means they they use this stuff so much no matter they might not understand how it's built but so if you have somebody that's digitally native that was born with it how would you then explain all the stuff we just discussed the language thing again or maybe Sushi making or where would you go I mean I don't know to be honest I mean I would try and figure out an energy an analogy of some description um the um you know is it is it making a car a modular car where you can pick and choose the um the doors and the windows and the engine and the and the batteries or whatever that could be one way of of saying it you know if it's if it's um monolithic you as an individual you can't touch it it's it's closed but if you if you're um if it's truly composable then it's fixable you can go in and change it you can go in and and take out the broken engine and put a new one in you can go and take out the the um the broken spring and fit and put a new one in depending on how granular your modules within the car is you know what's interesting to me is the more modern the card the more monolithic it is try to do this within Tesla you're not gonna take that apart and let's we could we could go into a very political oh boy Rabbit Hole here as well I think the the way technology is going is it's it's becoming less and less uh yeah and then easy for people like you and me who grew up with technology taking taking things apart putting them back together to figure out how they work we had the ability to still do that yeah exactly so you you say 15 year old who's digitally native yes they're digitally native with what they're exposed to they don't know how it works true I don't know how it works behind the scenes if it breaks can't fix it because they don't they they've never had to yeah exactly put that together again because you cannot do that these days you can't do that and so what I've learned from one of these I asked this question on one of the other episodes and the person came back and I think that's that's one that can work is actually remember these kids might have been old enough to see their parents with an old Nokia phone right it wasn't really composable it's like a monolithical phone it had a game and it had a phone like it had snake you know this stuff and now you can get like one of those guys where you get a base fund that can just call and browse the internet but then you can make it your own by choosing your games and whatever you wanted you don't like a game take another one you want more of that game pay a little bit maybe that one fits pretty well for a 15 year old yeah yeah I like that I like that and I think the um the the DHC um Arena just to just to sort of close it all off a little bit is is still very much experience focused as in how the assistants are presented to the outside world again to Alex's point that we we haven't really touched on how we're aggregating and pulling in the data further down stream so the moment we still point to point almost we still see very domain oriented it's very um Silo oriented you go and pull your your order data you go and pull your product data from a certain system you're going to pull your inventory data from a certain system and then aggregate them further up the stack why don't we do orchestration and composition further down this day that's what will happen next we'll have digital operations orchestration that's that's what it'll be but does that mean that there's more smartness of how to deal with that stuff down the channel let's say in a front end of an app or is there a thing in between in between in between because you want to be able to you want to give your Ops staff you know people looking after Logistics people buying you know if you're if you're in a in a retail context you want to be able to buy from multiple different suppliers and you want to be able to put that information into a system and allow that system that data to percolate up into the experience how you get it from system age system beat system T because they all need different views of that data oh that's true so there's states that they care about in a certain context yes and so this is the thing where the focus with current DXE is well there's a few problems to solve and the biggest one right now is that content editors can just not deal with stuff somehow and it's hard to be able to connect to all the things from the front end that's only one context and you're going to have other channels like staff in a warehouse that has to see a certain data set it's a different experience isn't it are you going to do that in your DXE or is that maybe not built for that that's a great question I love it and I'm going to be asking my CEO about that um soon okay it makes it needs to be answered I think it's something it's a very interesting one be able there's like and this is what I hope the DXE thing will be is that it's as big or as small as you want it to be right maybe that DXE bit is only for your the three channels you have your mobile phone your you know your in-store screen and your website but then maybe your procurement department or in warehouses they might have some other thing the more you choose to put it everywhere because the one thing that's interesting on DXE is that you have all these different um you have information about all the different systems so you can do amazing insights about these systems so you kind of want it to be big but then maybe it's I don't know I'm just spitballing here because it's so new we haven't actually done it at that scale we have to figure it out I agree so um with that bombshell we're getting towards the end I think um we're going pretty long now um I have a few comments here I'll put on because this one is cool uh lyron says it's it's like Roblox which is a really cool game for youngsters I haven't played it but as far as I know in Roblox you can make sub games and these games are exactly how you want them to be and these kids make the games it's insanely cool also politically dangerous but that's another story I I have I wasn't going to go into that but I have heard that too yeah yeah exactly and then one last one here from an anonymous LinkedIn user the human body is a perfect example the body is not built monolithic but he's composed of several microservices the brain orchestrates all parts that took it really far and really awesome I never thought about it that way it's true the housing then decompose and recompose the human body that's a little um too much composing for me [Laughter] yeah is there an a last thing you might want to share with our audience different age groups so five year olds and that's cool and so cool so it kind of um attempts to to do that and that's really the composability isn't going to be for everyone the the the customers that have a very um you know defined uh ammo with the contract with the outside world don't necessarily want to be able to pick and choose and change uh everything they want to get out there and get get their brands marketed so it really is important to understand the complexities understand the the downsides of a you know a truly composable architecture before you embark on this don't just go and Chase and you shine I think yeah because it's in the nerve behind in the boardroom we have to go that that might not be the real reason you have to actually do it yeah so go go in with eyes open and understand your business processes before you try and go confusing yeah and you know what what's actually interesting to me is there's there's quite a lot of systems now that are actually headless and composable but they still have part monolithical and they still exists it's like a surface and for example Bigcommerce like I don't want to name anyone specifically but they are such a good example because they have this whole headless thing and then a lot of companies don't want to build their own checkout and Card Services you use their monoliths for the checkout and you don't have to worry about the user Journeys oh man I would use that it in any second it's perfect yeah yeah I completely agree with it yeah um thank you so much um where can people find you if you want to be found that is um I'm on I'm on LinkedIn a little bit on Twitter okay my handles on the uh on the name tag type in Dominic and they'll find you yeah they'll find it maybe yeah we are I'm I'm relatively vocal on LinkedIn these days yes given my opinion to anybody who have it and you'll obviously find me at e2x and apply digital um in in London and Canada and maybe in Zurich you know exactly um thank you so much for your time this was good fun cheers