#132 – Enterprise Architecture: Selling Options for the Future, at a Cost – with Gregor Hohpe

Podcast Notes

In this episode, we succeeded to bring in Gregor’s perspective on architecture as a strategic, systems-level discipline not as a technical practice.

Drawing on his experience as a long-time advisor to large organisations navigating platform transitions, we explore how to bridge strategy and implementation, and how to create coherence across silos, enabling teams to make better decisions together.

Join us as we discuss how architecture guides strategic choices and helps you build optionality.

Key highlights

👉 Architecture can also be seen as a practice of selling options – enabling organisations to defer decisions and adapt as strategy and context evolve.

👉 Optionality always comes with a cost: more flexibility introduces greater complexity, so architects must continually balance benefits against trade-offs.

👉 Good architecture cannot be designed inside IT alone – it must be grounded in business intent, market direction, and strategic positioning.

👉 Domain understanding shouldn’t live only with data teams – it requires joint meaning-making across business, tech, and architecture.

👉 GenAI amplifies organisational dysfunction rather than fixing it – faster code and automation expose weak strategy, unclear domains, and siloed thinking.

👉 Those who work only within narrow silos are the most replaceable; future-relevant capability lies in boundary-spanning, systems thinking, and cross-domain judgment.

👉 As technology accelerates delivery, organisations must strengthen reflection, modelling, and decision-making – because the bottleneck shifts from building software to understanding what to build.

👉 Shared ontologies and domain modeling are essential for collaboration and extensibility – without them, organisations struggle to integrate partners, ecosystems, and platforms.

Topics (chapters):

00:00 Enterprise Architecture: selling Options for the Future, at a Cost – INTRO

01:32 Introducing Gregor

02:36 Beyond Business vs Tech

07:06 How does organization attitude connect to its architecture?

15:42 Designing Architecture for Strategic Coherence

23:01 Creating Shared Ontologies

29:53 Changing the Narrative from Transactional to Conversational

43:17 How can organisations remain context-conscious?

50:24 Breadcrumbs and Suggestions

Simone Cicero

Yes. And yeah, mean, think technology evolution is pushing them to do it. And also they will have no much choice coming up, I guess. That’s my impression. I mean, I wanted to maybe add a last bit, but since we don’t have much time, maybe more like a reflection that you can resonate with and then we move ahead with the breadcrumbs. 

So I was thinking that…

This space we’ve been talking about, domain understanding and domain modeling, as a standard piece of organizational development that sometimes gets lost between the IT or technology and the business, and that the data teams often look into from a too technological side, I think sounds a lot like what in AI development we are starting to call context engineering. So the idea that we have to understand the context, right? And what we discovered is that AI, GenAI needs context, right? So for me, it sounds like it’s not that AI, GenAI needs context, it’s that everybody needs context. And context is very important to share decisions and trade-offs, like you said.

But the fascinating thing is that the emergence of these AI systems, agentic systems, or even development systems that can leverage on the context is kind of showing us that there’s a lot of biases and pride and alibis and depth that we try to protect in the organization. And then suddenly there’s this technology that needs a clear understanding of trade-offs to help us take decisions. And it’s kind of nudging us into getting our organizations much more real. 

So what is your impression in the work that you are doing with companies as these technologies come up? Is it something that resonates with you?

Gregor 

Oh yeah, big time. And I see sort of two parts. The first warning I would give is if you’re living in a silo, you say like, only do my thing and I care less about the other stuff, you make yourself a great target for GenAI replacement because it works extremely well in silos. If all you do is like, hey, give me the requirements and I code them out, well, LLMs are pretty good at that. 

If all you do is design beautiful user interfaces without caring about the business domain. GenAi is pretty good at that, right? So that’s my word of warning. If you live in a little box in your organization, think again, because you are the prime target for replacement, right? Where AI is not as good as strategy, collaboration, looking across the different domains, dot connecting, taking bets, placing bets for your business. That’s where it’s not nearly as good. So folks who cut across, think, will be much more in demand. 

It’s just like everything else in the past. GenAI raises the bar. And the bar usually goes up by understanding the domain, working across the different silos. And that’s where I think you don’t need to worry about it. It’ll be a booster for you. But if you happily live in your little silo, it might actually be the replacement. So that’s big thought number one. 

Big thought number two is, it’s funny how organizations always believe that the technology will solve all their problems. And you can understand why. Otherwise, it’s a little depressing. But here’s the reality is a lot of new technology doesn’t solve your problems. It actually highlights your problems more. It amplifies the problems that you already have. 

Let’s start with something simple, the cloud. So it’s basically, if you have this giant separation between development and operations, because it’s not like silos just exist between business and IT, the silos also exist within IT. It just keeps going. And then you go to the cloud – Companies have taken a decade and are still dealing with sorting out that nightmare. How do I get the release cadence, the agility, all the things that the cloud provider promised me didn’t really happen because of my own organizational dysfunctions? 

And I think the same thing is true with GenAi. 

So I’ll give you a couple of examples. Let’s, for argument’s sake, assume that GenAi will actually make developers five times more productive. 

My prediction is, everything in your organisation is going to fall apart. Because the friction you have, the decision-making processes, the release processes, basically nothing in your organisation is designed to deal with a five times more productive organisation. Or the other example is what Simona alludes to, is if GenAI can build software just on a push button, let’s say you just give your thoughts, and the software comes out, for argument’s sake.

Well, you need to have pretty good thoughts, right? Because everything else has now become a commodity. So you need to understand your domain, the ecosystem, the language, all the things that Simon referred to, right? That is now the real differentiators. And in many organisations, that skill is weak. IT and business interface is the business pours money and wishes in, and IT sort of spits something out which may or may not be right.

Simone Cicero 

You know what’s funny? What’s funny? Sorry for interrupting you, but I think it was fun. It’s funny to chime in because, you know, at Boundaryless we have been thinking about some software we wanted to develop and for some months we have been a bit frustrated by the need to raise capital and so on. But then suddenly there are these tools that can help us prototype. And now I’m even more frustrated because I don’t have enough time to work on the domain modeling. so the replete agent is there waiting for me since a few weeks and I’m so frustrated. 

So if anything, the frustration has grown with these new technologies.

Gregor 

So you know it.

Gregor 

Yeah, and then multiply this by sort of 1000, and then you see how this goes in large organizations. I actually love your story because it tells us that we can see the effects in the small, right? And then, you know, for organizations, we can extrapolate. But that’s what I predict a huge challenge to be. You can no longer hide between, or you can no longer hide behind the long delivery cycles and all the specifications they need to write. You need to now go understand your domain, look for business opportunities, and build ecosystems, right? And that is hard work. I can totally relate to what Simone is saying, getting the bandwidth or getting focus time to now do real critical thinking to feed your AI agent in a way is actually harder than sitting down and just cranking a few lines of code or spewing out some random design document, the bar actually goes up. 

And if you’re not good at this, you’re not going to be happy with GenAI. So it amplifies the dysfunctions that you already have rather than making them go away. And I would say that’s the main warning I have for organisations.

Shruthi Prakash

So very curious how all of this shapes out. And yeah, it was really nice, Gregor, having you. And before we wrap up, we always love to leave our listeners with maybe a few breadcrumbs. So if you have, let’s say any books, podcasts or thinkers you are influenced by or have shaped your perspectives lately, it would be, you know, good to go into those.

Gregor 

Maybe I’ll start with a sort of meta comment. One thing I learned when I attend events, which I hope many listeners will do, I highly recommend for people to attend talks from people whose point of view they don’t agree with.

Listen to people who have a different point of view because one of the biggest trap I see people fall into, and social media doesn’t help with that, is that people look for things that sort of reconfirm stuff that they already believe. Look, I was right. This Gregor guy says the same thing. Platforms are important and the cloud. People look for reconfirmation. Instead, you should look for a different point of view.

I remember well, I went to one event and somebody, was working at AWS in the serverless department, and a very smart gentleman gave a talk, was basically, serverless is outdated, it’s based on assumptions from 10 years ago, and basically, it’s a bad idea. I’m like, that’s the talk I want to go to, because I want to hear what this guy has to say. And it was very, insightful. So that would be my meta tip.

In terms of concrete resources, I think if anybody hasn’t realized that systems thinking is now no longer a niche but a prerequisite, then take my word for it. So anything you can read around, whether this is the fifth discipline or thinking in systems, many, many related works, I think that is now required reading. There’s the domain modeling.

I’m sure every audience or every listener has already understood domain modeling and domain driven design. But then there’s also the systemic thinking, the systems thinking. So I would highly encourage people to, as I said, no longer consider that a niche, but required reading. And then perhaps allowing myself to be tiny bit biased, but it fits well within our podcast as well.

I think platforms are also no longer something that the big hyperscalers do only, right? It is something that no matter what business you’re in, you’re gonna be a platform for something else. So that’s why I wrote Platform Strategy. Quite honestly, it wasn’t an easy book to write because it has so many nuances between marketplaces and developer platforms and cloud platforms, all these kind of things. 

But I would also encourage people, it’s hard to have an IT strategy today that does, or even business, business slash IT strategy, let’s put it this way, without a thought of platforming it. So this would be maybe three vectors for our audience that should be useful.

Simone Cicero 

And of course, if you want to add a couple of bits of where the audience can catch up with your work.

Gregor 

So most of my work I have at architectelevator.com. The architect elevator is about seeing architects as a connecting element across the different floors. I would say from the penthouse to the engine room and back because that’s a lot of the disconnect that we just talked about. If the business has some vision in the abstract and the techie people play with random tech stuff, that is fun, but it’s not a good way of organizing your business. 

So your architectelevator.com, that’s where all my blog, my books are. And the only thing that readers need to keep in mind, I write the elevator myself. So you might read one blog post about serverless asynchronous messaging architectures, and then you read another blog post about platform strategy. But my take would be that’s actually good for you, because once again, you should be reading things that don’t just reconfirm what you already know but also take you into slightly different domains.

Simone Cicero 

You must dig the tunnel to the truth from both sides, right?

Gregor

Yep, exactly. Nicely put.

Simone Cicero 

Good, good. mean, was an amazing conversation. I hope you enjoyed it also.

Gregor 

Yeah, no, very nice conversation and great insight for questions. I was only half joking when I said the first question could easily take us an hour to answer. And I think the same is true for all the other topics. 

Simone Cicero 

But to some extent, we have been wrangling on this conversation for the whole podcast. So I think that was effectively what happened. It took us one hour to just wrangle around it. So thank you so much again. It was a pleasure to have you. I think it was a long due conversation for your work in platforms. It’s so relevant that it’s strange that we are having it at the seventh edition of the podcast event series. 

But for our listeners, of course, you can head to our website. You go to boundaryless.io. If you’re listening to this podcast now, will find Gregor’s transcript with all the links, the books, the things that he mentioned in the conversation. And thank you, Shruthi, so much for joining us.

Shruthi Prakash 

Thank you. Thank you, Gregor. Thanks, Simone.

Simone Cicero 

And yeah, until we speak again to our listeners, remember to think Boundaryless.