Low Down on Low-Code

Summary
Intrexx CTO Robert Hutter joins hosts John Rymer and Dave Marcus to discuss why AI is changing how software gets built in conjunction with low-code platforms. He explains how governance, process ownership, interoperability, and human-agent coordination help enterprises turn fast prototypes into secure, lasting applications.
Key Topics
Low-code is about creating a shared working ground for developers, architects, QA, UX, and business users, not simply reducing code. App generation is low-code’s evolution: AI-assisted speed combined with governance, compliance, and control.
AI can accelerate application development, but Hutter says the essential questions are what problem the team is solving and how it retains control over the solution.
In regulated European environments, compliance, validation, testing, documentation, and certification can outweigh coding. One pharmaceutical customer estimates engineering at 10% of effort and validation-related work at 90%.
A platform strategy can make certification and reuse more efficient by keeping applications within a governed, certified context; sovereignty also includes control of processes, applications, and datasets.
AI agents should be assigned tasks within a designed business process, with attention to context, results, and collaboration among agents, people, and APIs.
Pricing will likely mix seat-based models for people with consumption-based models for AI services and agents; Hutter expects human interfaces and teamwork to remain important.

What is Low Down on Low-Code?

Join us weekly to discuss the latest and greatest in low-code and digital process automation with executives and experts. Real conversations, no marketing BS. Hosted by Rob Koplowitz, John Rymer, and Ryan Duguid. Visit analysis.tech to get in touch about your personal low-code journey and learn about ways we can help.

John Rymer (00:00)
Hello everyone and welcome to the latest edition of ⁓ the Low-Down podcast by analysis.tech. I'm one of your hosts today, John Rymer,. I'm one of the principals at analysis.tech, and I'm joined by my partner, colleague, and friend, Dave Marcus. Dave, welcome. How are you?

dave marcus (00:22)
I am good, John. ⁓ getting getting used to being back in the Pacific Northwest from ⁓ sunny South Africa. ⁓ so that's been kind of fun over the last few days. but otherwise I'm I'm all good and always happy to ⁓ host these sessions with you. ⁓ today we have a guest that ⁓ some of you may know, some of you may not know. ⁓ Robert Hutter is the CTO of Intrexx ⁓ Robert and I have actually known each other for a quite a long time.

John Rymer (00:28)
Yeah.

dave marcus (00:50)
back to my UiPath days when he was a CEO of a company called FireStart ⁓ in Europe. And so it's great to have Robert on ⁓ with us ⁓ today. ⁓ thanks Robert for joining us and ⁓ we hope that people find this informative. John, back to you.

John Rymer (01:08)
Okay, Robert. So the way we like to start our podcasts is to ask our guests to tell us about their journey. So how did you arrive at this point in your career ⁓ at Intrexx ⁓ tell us a little bit about that, please.

Robert Hutter (01:16)
Mm-hmm.

Sure. So first of all, thanks for the invitation. Super glad to meet you again after we've met in Germany on the local days. So it's a pleasure to be here. Well, where did it start? so actually I'm a professional soft engineer from my educational background. And after graduation, started my first startup in the scene of banking application development. So very

Needy greedy critical stuff to build for large-scale banks in the portfolio management, asset management area. And after that, as you mentioned, I started my second company called FireStart in the space of process management automation, kind of bridging the gap between people designing processes and people actually implementing processes, and kind of do the lost in translation effects, get rid of it and make it more smooth in the transition.

⁓ also then the ending up in the whole intelligent automation space, so coming from RPA on the one spectrum to process mining intelligence on the other end. And we were kind of a bridging technology in between with this translational functionalities we served from process design to execution. ⁓ And what I've learned in that phase was always like ⁓ process is always the the critical part in how effective you get your business up and running.

But you also have a lot of dependencies, like to the applications that you're connecting, to the data sources you're connecting. And whenever there's something outside of your scope moving, your role is like the one to be blamed that your process is broken and not moving again. Which eventually kind of made it interesting for me to join InterX in in the CTO role, because here we have a broader scope on capabilities that we serve on the platform.

Not just the process automation and orchestration part, but also the data governance and application development part. So this makes it a lot more comfortable to build large scale enterprise applications that are more independently run and built from outside ⁓ factors. And this is a essentially what we do here at Intrexx, serving as a platform for companies building dozens, hundreds of critical applications on one platform strategy.

John Rymer (03:40)
Robert, tell us ⁓ something personal about yourself that a lot of people may not know.

Robert Hutter (03:47)
⁓ so actually I'm I'm a very curious guy in many ways. So I'm interested in sports, music. I was a gamer back then, so played also a lot of video games ⁓ originally. ⁓ but I'm also like very a drummer myself, so not just listening to music, but ⁓ to drumming myself, played in very different bands ⁓ from ⁓ different styles of music, and I think there's a lot of analogy also between like

setting a process is like on the drumming, setting the the the the speed and and if you're free jazz or if you're solid rock, you know, ⁓ the the tempo you give to the business always ends up how can people kick into the tune or not? And this is like a lot of analogy to process design. And how well designed is the process that people can kick in in their day to day business life and and be an added value to the whole business that you're running.

John Rymer (04:21)
It's hard.

Robert Hutter (04:46)
So that's also something that always fascinated me, some sort, yeah.

John Rymer (04:48)
Interesting. You're not the you're not the first person who has ⁓ who has made a connection between software and music, Robert. It's very interesting.

Robert Hutter (04:56)
Yeah. There's like one fan fact is like BPM is also beats per minute. Yeah, so that's the the common line.

John Rymer (05:03)
There you go. There you go. Yeah, yeah. Thank you. Dave, over to you.

dave marcus (05:04)
Yeah.

Sure. So Robert, as we talked about ⁓ earlier, ⁓ we had the opportunity to chat with you six to eight weeks ago ⁓ at the Low Code Association conference in Hanover, ⁓ in Germany. And one of the things ⁓ that ⁓ you mentioned that you were fairly bullish ⁓ that as valuable as AI is, it won't kill, you know, traditional approaches to low code process applications.

⁓ can you talk a bit more about that? You know, why you said that and how you think about it?

Robert Hutter (05:43)
I mean, from my perspective, the whole idea on low code is kind of bridging the gap on the collaboration side between professional developers and kind of citizen business oriented developers. and the thing is not about the code, it's about building a common ground that people can collaborate together in various roles on building a solution. Because the reality is not it's it's not like a single developer job to build a complex solution.

You have architects, you have quality assurance, you have UX design. So there are a variable variety of functions from very technical to non-technical. And from my perspective, the whole idea of low code is build a common ground that all of these people can kick in at a certain point of time, engineering a professional enterprise create solution that can last. ⁓

But the component level and the model-driven approach. So that you're not always starting from scratch building code on the very slowest end, but you have high level of reusable, secure, governed controls and components that makes it easier to compose faster in delivering the solution to production. Because that that is what matters. Like if you have a if you're facing a problem, how fast can I address that problem with a proper solution?

build on my platform. That's I think the core idea behind it. And does it take me two days or half a year is a critical factor in how fast can I respond to these problems. And I think like I'm an engineer myself, but I think people, especially these days, are too much obsessed about code and not enough obsessed about what is the actual problem we want to build a solution for. And I think this is where the whole game shifts because the coding part

is more and more handed over to AI assistants or services, but actually what is the problem we are addressing and what is the proper solution we need to deliver, this becomes more and more critical now from a perspective.

dave marcus (07:53)
Got it. John, over to you.

John Rymer (07:54)
Awesome. Robert, while we were in while we were at the conference, we heard a a wide variety, heard about a wide variety of experiences using AI to create business solutions. ⁓ There were some quite difficult requirements that that we heard about, ⁓ sovereignty being the one that really stuck with me. and ⁓ and modernization.

we heard a lot about the new platforms that say the government agency, the governments in Germany are are wrestling with. ⁓ it it it in in those mission critical applications, ⁓ therefore it's AI is moving along a little more slowly, perhaps. But I wonder because you guys are right in this right in the middle of real serious enterprise applications.

Robert Hutter (08:32)
Yeah.

John Rymer (08:51)
How are you seeing your European customers and prospects using AI alongside of the Intrexx low code platform? What kinds of things are they doing?

Robert Hutter (09:03)
so we mostly kind of introduce AI assisted services on the design time and engineering support, because this is where you can have most impact on speeding up the whole the application development process. But I think the the speciality in Europe is ⁓ we have ⁓ of course a ton of regulations that we need to fulfill ⁓ from a legal standpoint, compliance standpoint. So we always have

John Rymer (09:11)
Mm-hmm.

Robert Hutter (09:30)
To keep in mind in our design and our platform capabilities to meet these regulations, whether that's the AI Act or the Data Act or Accessibility Act, which is from my perspective not necessarily a bad thing. That that because like speed is nothing without control, in my perspective. Yeah. So just like rush rushing down the street 300 miles per hour with a wire coding fast car, you'll probably not survive the next curve around the corner.

Yeah. So it's about like speeding up, but in a manner that you can always stay in control. And I think like the governance aspect of staying in control is the factor that counts most these days. Because once you lose control on your code, it's gone, it will not come back. So like besides like speeding up the development process, how do we can how do how do we keep the control is like the central question we we discuss every day in our engineering teams.

Because that's one of the most the biggest value of our platform. If it's a trusted platform that keeps you in the driver's seat and makes you stay in control in what you're building, that is the greatest value we can deliver on the platform side. And coming back to the regulations, ⁓ just recently talked to a customer of ours in a highly regulated pharmaceutical environment. They say actually, like on the whole engineering effort.

Just 10% is the actual engineering or coding. Ninety percent is validation, testing, documentation, certification. And that's like because of the reality of the environment and the industry that they are doing. But the the the cool thing is like if they would have it built twenty apps outside of a platform strategy, they would have to redo do that for every application.

They say the beauty of the platform strategy with Intrix is we just have to certify the platform. But everything we build on top is within that certified context. And that makes the whole day-to-day business for them a lot leaner, a lot more efficient than without a platform strategy. And again, sovereignty is not just about infrastructure or data center. It's also about how can I own my process? How can I own my application, my data sets?

John Rymer (11:40)
Right, right. Excellent.

Robert Hutter (11:52)
So not just on the infrastructure level, but this is also sovereignty on can I stay in control of what my process is doing? ⁓ and that's also a question we try to solve ⁓ on an application and solution level with our platforms.

John Rymer (12:06)
Right. Thank you. Hey ⁓ folks, if you're for those of you who are still listening or watching, you are you are listening and watching to the Low-Down from analys by analysis.tech, a mostly weekly podcast about AI, low code, application development, and all the issues and opportunities around that. analysis.tech is an advisory

and a content creator and a thought leader. I'm one of the principals. Dave is another one of the principals. Rob Koplowitz, one of our our ⁓ our process experts is is in the group. Francis Cardin, also ⁓ the he's the godfather the father of of ⁓ RPA is in the group. ⁓ Rick Greenwald, our data guy. If you want to engage with us, ⁓

hit us up on analysis.tech. It's that simple. Thanks for joining. And now we'll continue with our conversation with Robert. Over to you, Dave.

dave marcus (13:11)
Sure. So Robert, you talked about this ⁓ a little a few minutes ago, ⁓ but certainly at the conference as well, there was a big focus around this idea, ⁓ John calls it the control plane ⁓ or a governance platform. And I think that you know, for AI generated software, ⁓ how do you see can you elaborate a little more around how you see this working for Intrexx customers specifically? and do customers have to sort of evolve their governance strategy

in order to fit into the way that you're delivering these capabilities, you know, how does that change? You know, we're seeing lots of scenarios right now where, to your point, you know, people can build really quickly and really easily, but getting to a point where you can build with control is a whole different animal completely. And so, you know, can you talk a little bit more about that?

Robert Hutter (14:02)
Yeah. So certainly there's like the technical aspect on it. So like what are the capabilities we can support on the platform to make it easier? That starts with security, endpoint security. It starts with reusability. Like do I build for one application? Can I reuse a data model? Can I reuse a process? Reusability is a very big part in how lean, effective can I build my solutions along the way.

But of course, there's also like this organizational aspects. And I think the main part in that regard is ownership. Like with my history of process management, establishing process owners was always the hardest organizational part. Because people were trained on hierarchies and not on cross-functional processes. So anybody was ⁓ scared to hell when I told, Hey Dave, you're now the process owner from Procure to Pay, have fun with

aligning all the different interest groups along the process. And I think that is still a topic not that well established as it should be. And now that comes more on the data part as well. So data governance comes along with data ownership. Who is owning the data set, who needs to approve changes on the data interfaces. And like managing all of these dependencies with clear ownership.

Is one big portion to make a platform strategy work, not only technically, but also organizationally integrated in people owning their domain and their data sets and their processes. And and that's like what we like to facilitate. And the better the functionality of the platform, the easier it is for that organization to adopt with that functionality and establish these roles and principles within the organization.

John Rymer (15:52)
Do you have do

dave marcus (15:52)
Got it.

Robert Hutter (15:53)
And now I mean today if you look at like this whole of course like white coding, let's say hype that comes from the the personal usage. Yeah, people got excited, non technical people got excited to get built prototypes very fast. But you have to differentiate differentiate is this like a personal prototype in building, or is this an enterprise grade application used by thousands of people along twenty other applications? That's a different game.

dave marcus (16:04)
Yeah.

Robert Hutter (16:22)
And I think this is now where, like coming back to the initial question, what's the future of low code? My perspective it's app gen. This is the transition that we did with Intrexx, coming from all of these basic principles on low code, and now transferred that with AI support into app gen platforms that give you the same usability, the same speed of development, but always with this governance and compliance and control aspect in the backyard.

John Rymer (16:49)
I I love the I love the fact that you you went quickly to the organizational ⁓ changes and the organizational discipline that that people need to think through. Have you has has Intrexx published the advice that it gives to to customers about how to set up or evolve their their organizations, Robert?

Robert Hutter (17:13)
Yeah, sure. So we have an own consulting team who delivers best practices along the the platform implementations. But of course we also have a lot of specialized industry specific partners that can can really establish in their verticals ⁓ these these principles they need to have on the organizational side. Yeah. And it's always the problem usually is not the technology. The problem is usually the way you address the technology from the organizational aspects, and that's still

a tough topic that takes some time, you know. This is nothing you can turn around overnight, ⁓ especially with the speed that you see today with AI development, you know, this this goes crazy. ⁓ and you know like people are still like shuffling their emails around like 10 years ago, but they are now overwhelmed with these capabilities. And this is especially you should not should not just throw yourself into this madness. You should like adopt to it.

but at a certain speed on the control that you can digest with your organizational teams that you have. that's my advice to

John Rymer (18:20)
Right. Right. All right, let's move on. ⁓ we're we're gonna talk about technology again. Sorry, 'cause I agree with you, it's not the hard part. But one of the themes that came through in our conversations in Germany, ⁓ Robert, was ⁓ a desire by many for openness, for a level of openness in in the low code among the low code platforms that that was really expanded. So people wanted to be able to use

Robert Hutter (18:28)
Of course.

John Rymer (18:49)
say Claude Code ⁓ to drive application development on a platform like Intrexx. ⁓ and then others are very interested in in basically opening up the back end so that the a low-code platform can feed their DevOps pipelines. ⁓ I mean there's lots of ideas out there. ⁓ what did what do you what is your thinking about this?

Robert Hutter (19:16)
Yeah. So

John Rymer (19:17)
Is it is it a good thing? Is it a bad thing to be more open, make it more complex? Or is it a good thing?

Robert Hutter (19:22)
So so first of all, we are open as anybody can be in terms of APIs standards. So I'm a big fan of interoperability and standardization because that makes platforms also work together. ⁓ but it openness doesn't mean open source in my opinion. Yeah. Because we like we are not acting on on source code, we're acting on components and APIs.

dave marcus (19:43)
Yeah.

Robert Hutter (19:49)
And on that level, we're super supportive from open open API, open telemetry standard, ⁓ BPMN standard. So there's like a dozens of standards that we support with our APIs ⁓ to be very flexible in the way we integrate with other solutions. But again, coming back to like if we want to provide governance and security as a platform, we cannot AI allow to vomit

John Rymer (20:08)
Mm-hmm.

Robert Hutter (20:18)
uncontrolled code in our platform. So this is where we are protective, you know? ⁓ and this is but this is the benefit of the platform because we make sure that everything that happens within our platform is in control, is governed, is secure. And we talk in the backyard with cloud code and open AP ⁓ Chat GPT to use these services but then do the back translation on our secure component side. And that's again not

one vs. or the other, but it's like we're acting on solution engineering, we're not acting on code levels. This is why we don't need to open up our code, we need to open up our APIs and this is what we do very, very actively. Yeah.

dave marcus (21:01)
And and and and Robert, do you see the same thing applying with the role that Intrexx plays for you know, we use this word orchestration a lot, but the sort of coordination of not just third party data and applications, but also third party agents. You know, we're seeing this a lot right now where you know people are gonna buy agents from whoever the technologies are that they use. It's not like every single agent's gonna be built with a single platform.

Robert Hutter (21:10)
Yeah.

Yeah.

dave marcus (21:31)
And so, how do you think about the Intrexx's role with regard to coordination, management and governance of agents as well?

Robert Hutter (21:31)
Yeah.

Sure. I mean this is the part that happens mostly in our process builder and engine. And again, this is a a process problem. Because once you bring in agents, the first question is like, where do the agents fit? Which process, which tasks can they take over? And if there's a void on the process side, it's a question you're you cannot really answer. So again, ⁓ agents from a behavioral standpoint is like the the same problem like with people.

if I bring in Dave and say, "Dave you're super good with those skills," I need to figure out which team can I make you fit, which tasks can I hand to you? over. But this is again more of a classic organizational team setup problem, process problem, not necessarily a technical problem. But that that's ⁓ a very important topic again on the processor side, that when we orchestrate an enter and business process

We need to define is this a task that goes to the API, is this a task going to a human, is this a task going to an agent? But again, make them collaborate is part of the process design, how well they fit together in streamlining the whole process. And in in that regard, ⁓ I think the whole like one topic I I think that doesn't get enough attention from my understanding is task management.

dave marcus (22:49)
Yep. So

Robert Hutter (23:00)
Because the task is the very core asset, you need to think about context, accessibility, ⁓ what is the result, how do we process the results, and and do we give it to humans or agents? That's one of the core questions right now. So everything we're building is is like making the right context so that the task can be performed in in a very efficient way, regardless who performs the tasks.

dave marcus (23:15)
Yeah, I think it's specifically

Yep. I think it's specifically true ⁓ you know, in today's world where so much application of these sort of new technologies are being applied to automated tasking. But that automated tasking doesn't necessarily come with a context. You know, it's I'm just automating this task and I'm doing it selfishly because it's my task that I need to get better. But the whole idea of how it fits into something bigger is not necessarily well understood or well coordinated.

Robert Hutter (23:48)
Yeah.

Exactly. So yeah I also have to separate personal automation again from enterprise process orchestration. One is like I'm building my personal flavor automations, the other one is I have to design a business process crossing ten systems, ⁓ interacting with fifty people and making that robust and lasting. And that's a very again a very different architecture game that you need to bring in place. Yeah.

John Rymer (23:59)
Right. Right.

dave marcus (24:18)
So, Robert, let's talk a little bit about the business side of this. ⁓ you know, people have talked about the fact that, you know, traditional approaches to buying low code, and I'm just gonna use this low code or process automation platform technology, ⁓ you know, historically has become expensive, it's rigid. Do you see that the business models for both these and other technologies are going to have to change? And what does it mean for you guys?

Robert Hutter (24:21)
Yeah.

So first of all, that might be true for our US colleagues and competitors, but not necessarily for us. So I think we have a very feasible commercial model and also very clear predictive pricing model, seed based mostly. ⁓ so there's not there's not like surprises that would typically come with maybe transaction based models or asset based models. I've seen a lot of very confusing pricing models. So I don't think that we we do that.

⁓ and again the the beauty of a platform strategy comes with the interest effect. So the more applications you're building, the more it pays off economically. ⁓ because if you're having twenty isolated SaaS applications, you're paying 10, 20, 30 euros for each of them, and versus you have one license that you can reuse across multiple use cases. So that pays off very quickly.

once you have the platform available and the first use cases up and running, every use case you're adding pays off even more in an interest effect. And that's from my perspective, I would not see like what you said initially, that this is very true that might apply to some vendors in the market, maybe, but but usually the economic effect and impact is very positive with local platforms once they're established.

John Rymer (26:10)
Do you think we're do you think we're headed to consumption based? ⁓ with you know, with AI, it's all based on tokens and tokens are the are the key metric. does that push us toward that plus cloud push us more towards consumption based?

dave marcus (26:10)
Got it.

Robert Hutter (26:19)
Yeah.

Well ⁓

Sure. I mean like on the AI usage, it certainly will be some sort of consumption based models. But again, it's about if you want to make agents and people collaborate, people will be seed-based mostly. Agents might be consumption based. Again, you need to have just like different pricing models to fit the use case. Yeah. And I think the hard shift we've seen at the beginning of the year, where everybody was like, okay, we just kill the user interface because agents will take over the world.

This is not gonna happen. There will be some parts taken over by agents, but again, it's about making them collaborate with people, with teams. And this is not going to go away. So you will always have like a mixture of these pricing models, whether that fits for users or seats, or whether that fits for AI service integration, token based, prompt based, whatever, there will be a mixture of flavor from my understanding. Yep.

John Rymer (27:24)
So it's mixed it's a mixed model. Let's ⁓ let's saw it off there. Thank you so much, Robert. ⁓ p for people who don't know Intrexx want to know more, Intrex.com, I assume, they can visit and find ⁓ materials explaining what you guys do and the services. ⁓ I think the services you're providing to help people evolve their organizations, ⁓ I would highlight those if I were you. ⁓ because that's really valuable stuff.

Anyway, thanks so much for joining us. It was a pleasure to sit and and ⁓ and talk with you in Germany, and and it's great that we've been able to bring that conversation to our our audience here at The Low-Down. Thanks everybody for joining, and we'll see you next time on the next episode of the Low-Down by Analysis.tech.