 Those of you, and I know there were some of you, because I recognize you. Those of you who were at our Edinburgh event a year ago, you would have heard the first talk inside the open group of ecosystems architecture and some work that's being done by a subgroup of our architecture forum on this topic. And in that time, that group has worked diligently and had a lot of great discussions along the way that you will hear about shortly. But I'm pleased to announce that, as of today, with Phil a little bit more. So just a quick introduction to Phil. Some of you will remember him. If you've heard him speak before, you'll definitely remember him. It's great to have you back, Phil. He is, for the past 21 years, Phil has worked for IBM as a technical architect. And he's been CTO, data ecosystems, and he's also vice president of the Academy of Technology between 2019 and 2022. And he, in his day job, he helps top 100 companies build really big data and analytics IT systems. So without further ado, because this is a great topic. Don't want to waste any time. Over to you, Chris. Thank you, Steve. Yes, I was just going to say welcome, Phil. How are things in Preston? Hi, guys. Can you hear me? Can we hear it? Can you see me? Yes, yes. OK, so we did publish the e-book just now. You can get to it on the open group library. But there will be a hard copy following shortly, OK? So Phil is the primary author here of the book. We've announced it. Why is it important, mate? So just before we dive into that, Chris, can I just make something clear? So thank you very much, Steve, for introducing me as an AI. In my case, the AI stands for artificial idiot. So at least we've set the level of expectation. And I also should compliment Andreas for what an absolutely brilliant session. But before we start, because we're going to touch on AI slightly, let's just reset the tone. So Andreas talked a lot about what could happen. As architecture, we should really only be interested in what can happen. So ecosystems is essentially about what we can do potentially with AI now. That's one of the things that this is about. And actually, the book does not talk about science fiction. It talks very much about science fact. So on that basis, just repeat your question, Chris, and then I'll dive into it properly. Sorry. Why is this book important, Phil? Right, because there are a couple of things going on. So as we, as architects, learn how to build bigger, better, faster, stronger systems, there are two consequences of that. The first thing is that our systems scale. And the other reason is that as we scale our systems, the side effect is generally that the systems become more complex. That causes a problem for us as humans in that we only have certain finite limits. We have capacity. And the systems that we're being asked to architect these days, in certain circumstances, are becoming just too large and too complex for us to handle. There's a second problem as well in that if you look at current state of the art, as in, first of all, we start to think about how we architect systems. And then we afforded ourselves the luxury of thinking about systems of systems. In other words, what we call enterprise architecture today. There's a problem with the idea of enterprise. The world that we're living into today, and predominantly there I'm speaking about the internet aid, means that most enterprises or organizations are afforded the opportunity for almost limitless connectivity beyond the bounds of what we would traditionally consider to be a business or enterprise. So a question that we asked as a group of architects over 15 years ago now was, how can we work in an environment like that? How can we abstract above the level of the enterprise into hyper-enterprise contexts? And how can we work with systems within problem contexts where it's almost impossible to understand where the boundaries of our enterprises or our systems are? In formal terms, what we're actually talking about is the idea of closure in IT architecture. So the primary question that we asked was, how can we potentially think about architecting systems in a world where it's entirely open? And if you speak to experts from other fields, they have a world for that type of open context. And the word that they use is ecosystem. So in an ecosystem context, we're dealing potentially with hundreds of thousands, if not millions of components that are at the mercy of the ebb and flow of the requirements around them. And they could be dynamically changing or evolving within side contexts with which we no longer have direct control as architects. How could we work in that hyper-enterprise world was the challenge that we set ourselves many, many years ago. And it's taken a lot of thinking to get to the point that we're at now. I hope that answers your question, Chris. All right, well, I'll reread the book and we'll see. I can tell you that when Paul, let me just name the authors here so that you know who we're talking about. The authors are Phil, Paul Homan, Neil Fishman and Rahul from Honda R&D Europe. And some of the contributors, Steve Nichols from DXC, Mark Dixon from the open group, Christopher Hines from Rolls-Royce and Stuart Weller from Rolls-Royce. When Paul asked me what I thought about the early drafts, I said to him, my head's exploded. All right, there's some incredible thought leadership in this document. So there's a storyline and a thread that's run through this field that you guys have worked on for the past year. Can you talk to us a little bit about the composition of the book and the storyline for it? Yeah, Saul, if you would look at the history, back in, actually, I was thinking about this, Chris, because I know you and I prepared the questions up front. In around 2001, 2002, there was a group of us were deeply interested in the idea of semantics in software engineering. And that actually translates into IT architecture as well. In other words, how can we understand the things that we're talking about, are tangible and real? How do we link those things together? And how do we bend and modify that thinking so we can think about genuine complexity, not just in IT systems, but in a number of systems that are around and associated with both science and technology. Social systems, for example, social media. And that eventually led onto the early work in graph databases. And eventually we got to the point where not only were we interested in modeling architecture, you got to remember that the fundamental substrate underneath architecture is essentially the representation of the formal representation of ideas. So that work eventually progressed almost hand in glove with advances in information retrieval, for example, and early AI, until eventually we got this weird confluence of ideas about 18 months ago, where you now see AI's like chat GPT and BARD and other instances. Now, what's interesting is if you were to look at the formal mathematics under those AI's, what you'll actually find is the same, are the same mathematics that we were thinking about decades ago now, with regards to the linking together or networking of ideas. Now, if we were to peel back the ideas of IT architecture, if we were to look at the semiformal schematics that we all trade in as a matter of almost reflex these days, then underlying those schematics, you've got the same mathematics. If we were to actually essentially undress the schematics and redress them in a way so that you could represent any idea or theme, so long as you could capture the semantics precisely, then essentially with the aid of artificial assistance. So in other words, if we can augment the capability of an architect using advanced tooling like large language models, then essentially that springboards out into a whole new universe of possibilities and ideas. So a colleague of mine inside IBM many, many years ago coined an informal unit of measure for complexity. He referred to the head fault. And by a head fault, what he meant was that if you can't fit all the ideas associated with a certain thing in your head, then by definition it's complex and you need multiple head faults to work on that. What's interesting is if you map together, if you bond together all the good practice that we developed thus far in the profession of IT architecture, and you use the underlying first principle mathematics and then you bring in the very, very powerful and profound capabilities that we now have in generative AI, generative AI and large language models, what it allows you to do is to cast out beyond the idea of an enterprise and work in the domain of dynamically system, dynamically changing problem spaces. So this is, if you like, I'm gonna say it a different way. When generative AI models like chat GPT arrived about 18 months ago, I said to those around me, look, this is a case of everybody off the bus, we need to start again. That's not true, actually. If you look at what we've done over the decades of architectural and software engineering progress, actually what it's done is it's naturally brought as the situation that we're in now, which is if we apply all the forethought, all the governance, all the rigor that we bring as architect engineers, then there's absolutely no reason why we should not engage with this next generation of technology to allow us to think about what we previously thought to be unthinkable. Now, Andreas was right. We need guide rules or guide rails, but actually the guide rails are already in place through the professional practice we already have. What you will find in this book is the bringing together of all the first principle ideas, all concepts around guide railing that we already have inside IT architecture. You bring together all the best practice that we've had, and hopefully you come up with something that will take us into what we're now calling the third golden age of IT architecture. Yeah, so Phil, can you just comment on what the three golden ages were that you're outlined in the book? Yeah, so I'm stealing this. So this actually came from Grady Butch. We were lucky enough to work with Grady a little bit on crafting the text. What Grady says is rather profound. I actually love this. He said the first golden age, he actually refers to the golden age as a software engineering, but he kindly agreed that we can transfer into IT architecture. So the first golden age of IT architecture was the age, if we wanted to first with the machine, entirely conform to their language and protocols. So we're talking about the age where we first started off with machine code, then we progressed to assembler and high level languages and so on and so forth. Then what we did was we hit the point where we had to talk to the machines directly and immediately understand. What we afforded ourselves was the capability to abstract slightly further and talk in terms of encapsulated ideas. Now that's what Grady refers to as the second age and there you find instances essentially of professional advance or thinking around design. So you get the idea of objects and encapsulation. You get the idea of the schematics and the graphical ways of communication that we naturally evolved as a community. And then we actually got to this tipping point which is very, very recent surprisingly for astonishingly profound certainly for me. So those who will be familiar with the work of Alan Turing will most probably remember the Turing test. In other words, can a machine fool you that it's human? Well, I think we're probably at a safe point where we can suggest that we've smashed the Turing test. And what that means is that actually the coin has been flipped from the first golden age and no longer do we need to talk to machines in languages that they can understand, rather they can talk to us in languages that we can understand. And what that means is that if we are careful, if we are professional and if we are precise, then what we can do is essentially we can work with machines as our assistant to allow us to augment our practice to move out into domains that we could not previously reach before. That's what we refer to as being ecosystems architecture or architecting in the hyper enterprise plane. So, Phil, could you perhaps suggest some ways in which the architect's profession might use generative AI going forward? Yeah, absolutely. So I was in a meeting a week and a half ago with a fellow architect and he passed a comment that I didn't quite believe at the time. He said, I'm talking to large language models and I'm getting them to produce schematics for me. And he mentioned a specific large language model. And I corrected him afterwards and I said, look, I think you'll find that that particular large language model only converses in terms of text. It can't produce schematics, it can't produce pictures. And he said, no, no, no, no, you're forgetting something, Phil. I can ask that large language model to output its answer in SVG. SVG is a text based file format. So if I want a picture of two dolphins running up a street wearing a soccer strip, it can do that for me. And I said to him, pass me the code. And actually before I'd finished the sentence, I was online already. And I'd asked one of the LLMs. I said, can you produce me a class diagram in a specific format that demonstrates the subtlety of a facade pattern? Within five seconds I'd taken the output from that LLM. I'd gone on to another tool online and I had anything that I could have ever drawn of many components, all evolving and close to real time. Visibly exciting and interesting. You had, your audio and video had frozen for a moment there. After you gave that example to us a week ago, I went and discussed this with some folks relative not to UML, but to the Archimate language. And we have, Archimate can be expressed in XML. Just go back about 30 seconds, Chris. The line's obviously failing, I'm sorry. Yeah, what I'm saying is that you were giving the example of the LLM and a rendering of a UML described facade pattern. And when we discussed that last week, I said, well, okay, we can produce XML definitions of Archimate through the exchange file format here at the Open Group. Would it be feasible to do that with a large language model, the Togaf standard, the Archimate standard, all that sort of thing? And the answer that came back from the several people I asked was, yeah, there's no reason we couldn't do that. So the example is if it's something that we can talk about and describe in a fairly formal way, some of these things can accelerate a lot of the work that we would typically do. It augment the work. Yeah, so can you hear me at the moment, Chris? Yes. And the moment you say yes, it freezes. AlphaGo was, have you still got me? Yes, we've still got you. It's just Murphy's dancing around the room at the moment. Right, so the key point is this isn't just about SVG or XML-friendly formats. It's about the capability of LLMs to communicate in any form of language that we deem to be relevant. So we're likely going to be moving into a world of dynamic languages that will be created within the moment. And that in itself of a whole universe of possibilities and creativity that we as an architect community can only, we're only just on the edge of imagining now. So this is a horizon that is rising at the same time. And this is work that absolutely beside a formal community of architects who are interested in the betterment of the profession. I can think of no better work, no better place than the open group to do that work. So that's really an invitation to the community and the industry a large to come and help flesh out this work and collaborate with the ecosystems architecture work group. Correct, given the pace of advancement in generative AI at the moment, almost certainly we will see progress in the tooling space. I know that you're already talking to one or two organizations who are playing with ideas in the ecosystem space and around generative AI. I think we're heading for an amazing spurt of inspiration and advance. The key point, so the point that Andreas made was if we keep this within the profession and we make our work professional, then all that we can expect is increased professionalism and better practice and capability. Again, I cannot emphasize this has to be about what we can and should do, not what we could or would do. Yeah, the ship has effectively sailed and we're in the middle of it. I think this book has actually laid down and clarified some things that are actually in flight already and just haven't been necessarily pulled all down in one place if I understand the reason you guys got together. Yeah, so this isn't particularly about anything new. It's about the cohesion of a collection of ideas that have been in play for a number of decades. What we've done is just updated those within the context of this post-uranian world. I would appeal, every single one of those who have contributed towards this book deserves a round of applause. I'm exceptionally proud of it. It's been some time in the coming. It just happens that it couldn't have happened within a better community. It's just great news, absolutely great news. Exciting times, my friend. Indeed, what I would like to do, Phil, is just hand you back over to Steve for Q&A, from the audience, and thank you for the contributions to all of the authors and the contributors. Thank you, as always. I bow before you. Thank you, Chris, and thanks, Phil. We're running a little behind here. I'll keep the questions brief, but I did, I do swear that, in fact, excuse me, I'll give this back because it's past code's gone off and somebody else needs to do that for me. Please, Andrew. I swear, when you first mentioned generative AI at the beginning of your talk, you actually said degenerative AI, which I thought was quite funny. Did I really? Well, it was because I declared myself as being an artificial idiot. I do apologize. No, so actually, I did see the questions on there. There's one I need a refresher about, but the first one was, well, it was along the lines of thank you for explaining where you've got to in the journey so far. What's next with this? I think the inevitability is that we're gonna see tooling quicker than even I could imagine. My expectation is once the community gets hold of this, we'll see incremental advance. Well, my guess is it will match progress with LLMs. I think it has to. How fast that is? I think we all have to stand back and watch to be truthful. I don't really wanna predict that type, but I think it would be foolish to predict, but fast. Faster than fast. Faster than fast, there we go. So the other question that came in is, do you think the open group should establish an open interoperability framework standard that would leverage AI for defining a specific solution for any pair of enterprises? I think the default answer has to be yes. I'm very passionate. I think now is the time for standards organizations to step up and take the lead. We are heading towards a brave new era. It's right and proper that organizations like the open group should be front and center. So the answer is yes, but I think that should only be one of a number of working groups that are spawned out of the open group. This is a leadership thing. Yeah, yeah. Phil, I appreciate the efforts of you and all the authors and I absolutely second the idea that that should deserve a round of applause, which you'll hear another one of in just a moment. But I've read this myself and I found it really thought provoking. I actually thoroughly enjoyed it and I know I'm on stage at an open group event, but I don't always find everything we publish quite as compelling as this. It really did get me thinking about this and the timing could not be more appropriate. So I'm delighted that we're doing this now and look forward to where it goes next. So can I just end on one thing? Can you help me fulfill a life's ambition? I'd like to say one thing. Of course. Back to Steve in the studio. I've always wanted to say that. Thanks, Phil. Appreciate it. Okay, folks. Yeah, well, here we are in the studio. We're gonna break from the studio for coffee, tea, whatever, outside. We, as you know, we're a few minutes late going in, so let's take 15 minutes and be back here in 15 minutes. We have some more great stuff to go through and I recommend downloading this e-book to everybody. See you soon.