Low-Code Won’t Replace You | Camunda’s Bernd Ruecker

0:00Hey, Andrew, have you ever woken up in the morning and thought to yourself, I would love to get out of bed today and help Devinterrupted make content. Well, Ben, I think that most days when I wake up. Yeah, well, that's great. But I also want to give our audience that opportunity. And we'll get to that in just a moment. But you've probably noticed some changes that are happening here at Dev Interrupted. You know, our goal really is just to provide the best coverage of topics that matter the most to our audience. And to that end, we've launched a new survey to gather feedback from you, our listeners.

0:38we want to know what an ideal Dev Interrupted looks like for you. And this will help us make the best content and let you influence the roadmap. But there's one extra bonus that I'm going to let Andrew share with everyone. That's right. There's a little bit more to this too. As part of adjusting our content to make sure that we hit all the marks you want to see, we want to invite you to share your opinions on topics like developer productivity, developer experience, adopting and scaling AI. And this becomes a chance to join the conversation with us and engineering leaders that join us on this podcast about the challenges facing software engineering teams in 2025.

1:20The survey that Ben alluded to includes a section for you to indicate that interest, but even if not, we still want to hear from you. We still want to know your opinion on Dev Interrupted. It's only going to take a minute to complete, So please check out the link in the show notes. And thanks in advance for sharing your opinion with us. Thanks, Andrew. And for our listeners out there, I'm Ben Lloyd Pearson, and I'm joined today by my co-host, Andrew Ziegler. Yeah, of course. It's always great to be here. Excited to dig into today's topics. What's top of mind for you? Yeah, so, you know, I feel like we're probably not going to get through a single episode this season without mentioning AI at least once.

2:02Too true. So, you know, the stories that have been catching my attention lately have been, you know, it seems like we're going through another round of major investments into AI infrastructure. So, you know, there's the one that made a lot of news was this new Stargate project where OpenAI has come together with a few companies to invest half a trillion dollars over four years. But we also have stories of like Google investing a billion dollars into Anthropic. Microsoft a few weeks ago announced an$80 billion investment into their AI data centers in 2025. And then Meta invested as part of the Databricks Series J, which is an AI startup.

2:42It wasn't disclosed how much Meta specifically invested, but it was a 10 billion round overall. So somewhere under that amount. The AI hype cycle is clearly not over. And it seems like what's really beginning to emerge is that the U.S. and China are the only two major world powers that are aggressively investing in AI infrastructure right now. And more specifically, they're investing in controlling and creating the AI chips because that infrastructure is useless without the chips they're built to support. And so we're going to see these world powers controlling that AI dominance by controlling the creation and production of that technology.

3:22Yeah, it's like who can build the most hardware and then put the most electricity behind it to train their models, you know. But it's almost, you know, what it almost feels like to me is like there's a bit of like an arms race that's emerging here that reminds me a lot of like the Manhattan Project, the space race. And in both of those situations, previously, the U.S. was the first nation to really reach the heights of success for both of those challenges. And it put the country in a place to dominate. So, I mean, when I put it in that framing, it really puts things into perspective about what's happening.

3:57And it really shows why the next few years even can be so critical. It's still early days. There's something new every day when we wake up in this kind of development. And while it reminds us now looking back on things like the Space Race and the Manhattan Project, you know, those projects themselves are even completely different from each other and a world full of unknowns and uncertainties every single day. And I think that's the reality of what we're currently experiencing. For example, the recent Chinese model DeepSeek R1, which uses pure reinforcement learning, is able to match OpenAI's O1 power at 95 % less of the cost.

4:34And so what that tells us is there's no moats yet in the world of AI. No one is secure in terms of developing the next leading thing. I mean, it's really hard to shake the feeling that there are just massive changes around the corner. I kind of feel like 2025 is going to be the year that we really start to see massively disruptive outcomes from that. I really feel like 2026 then becomes like the winter for AI where investments are just going to evaporate. Like you can't invest this kind of money into core infrastructure like this and not expect significant payouts on the other end. And speaking of this kind of massive disruptive outcomes from AI, in the last week, the world of SEO was turned upside down by changes from Google in an attempt to protect its market share from AI.

5:25This was an interesting story that I learned about, and perhaps if you're in close communication with your company's marketing team, you've heard about this too. But there were many changes from Google's end in the last week that resulted in temporary data blackouts on SEO tools and things that actually connected to Google to understand page rankings. For example, Shea Harrell, the senior director of SimilarWeb, shared a statement saying that Google implemented significant changes to how its systems handled automated interactions, and it required JavaScript to be enabled for search queries. And what happened is it took about five days for most of those tools to start working again, which if you're working at a company that's buying and investing in ad space and lives and dies by the world of SEO, that's an extremely stressful time.

6:10And I see this as a play from Google to protect its dominance in the world of search, which is slowly getting eaten away by technology like AI. For the first time since 2015, Google's global market share has dropped below 90%. And we know that at this stage, LLMs, when they're scraping websites and collecting content, they're largely not executing JavaScript. So this seems like a short-term change to shake off the scrapers and give Google a little more time to protect its search supremacy. Yeah, I've been saying for a couple of months now that search is dead, you know, and it's a little hyperbolic because obviously, you know, even myself, I still use it on basically a daily basis.

6:51But I also am starting to wonder, like, could we be witnessing the death of APIs that provide free access to data? So, you know, Twitter, Reddit, they've already made their APIs more restrictive. Specifically, you know, I know Reddit, it was specifically because of AI training concerns, but I imagine Twitter had similar concerns as well. And I've started to wonder, is Stack Overflow the next one? They still provide public data dumps, but how long is that going to happen, particularly considering that they now have a partnership with OpenAI? And personally, I really love how these GPT interfaces provide a search that really does a good job at connecting disparate points between multiple resources.

7:37And honestly, that's probably been the single feature in these tools that have accelerated my workflows more than anything else. But, you know, there's still so many bugs to iron out with this. I personally experienced this pretty recently where I'm not going to share the term I was searching for, but it was a very basic search for like, what is meditation as an example? Like that's not too dissimilar from what I was searching. And the result that I got from Google's AI was it basically focused in on this really toxic version of meditation and tried to convince me that I might have a mental disorder for asking.

8:12Asking to define this word, it blew my mind. But then it got even more shocking when I asked somebody close to me to do the exact same search. And the response was almost the exact opposite. It gave a very like clinical approach that was grounded in practical advice. And it went even as far to encourage this person to reinforce people who were engaging in meditation. Like I couldn't believe how it was shocking, honestly, how different those responses were. And, you know, you mentioned DeepSeek earlier. They've been kind of a hot entrant onto this market. You know, one of the issues with them is they actively censor anything that goes against like the Chinese Communist Party directives.

8:54And really every single LLM has censorship capabilities like that built into it because that's fundamentally how they're designed to make decisions. If you want to see what happens when these controls aren't in place, go look at what happened with Tay tweets, I think back in 2016 or something, where it just went off the rails and said just extraordinarily hateful and terrible things in a relatively short order. so you know now more than ever i feel like the algorithm that feeds you information is becoming more important than it ever has been even if it's always been important and google is a really great example of what i think is going to happen to a lot of companies like traditional conventional things that we've all normalized over the last you know one or two decades are now having the the paradigm just completely disrupted.

9:48So I think every company, every leader should be thinking about this challenge and if it's going to impact them in a similar way. I completely agree. And by the way, for our listeners, for those of you that are not already following us on Substack, everything that we've discussed today is in the Substack newsletter. So be sure to check it out and subscribe and follow us for more details on these topics. Continuing to the topic at hand today, I want to start by setting the stage a bit. Engineers have commonly treated applications and tools like low-code design, like the beginner's lane of software development.

10:27But what if the real power move is knowing when to step into that lane? Here's the thing about low-code. It's not beneath developers, but a force multipliers for rituals like automation. And after the break, today's guest, Bernd Ruecker of Kamunda, will talk about low-code and other automation workflows, and how they fit into an engineer's day-to-day.

10:53Join Linear B and Luca Rossi of Refactoring.fm for a workshop to learn the three key traits shared by highly successful engineering teams. This workshop uncovers a wealth of insights about how to make your engineering organization more successful. You'll learn how to take practical steps to ensure engineering is well-regarded among leadership and non-technical stakeholders, your engineers are satisfied with their development practices, and to more consistently ship projects on time. Check the show notes to RSVP your spot in tomorrow's workshop, and we'll see you there.

11:29Hey, everyone. I'm your host, Ben Lloyd-Pearson, Director of Developer Experience here at Lanier B, and I'm delighted today to be joined by Bernd Rucker, co-founder and chief technologist at Camunda. Bernd, thank you so much for coming on the podcast today. Thanks for having me, Ben. I'm delighted. Today, I want to spend some time just talking about something that seems to be pretty core to your professional experience, and that's process automation. You founded a company that is sort of centered around this concept. So before we dive too deep into it, because we have an audience that comes from all different types of backgrounds.

12:06Many of them are probably not super familiar with what process automation is. Maybe if you could just give us from your perspective, like just an overview of what process automation is and how you define it. I always like to do examples. We could pick basically an industry. At the moment, I'm working with a lot of banks. So let's just use something from there. For example, the bank account opening process. So if you want to open up a bank account, you basically go to the website of your bank, probably enter some data and then you move through a process where a lot of things have to happen in a certain sequence like probably you have to identify yourself then there are checks on the bank side like a fraud check and a trust check and know your customer compliance checks sanction checks whatever until you're at some point in time hopefully get a bank account so the bank account must be opened an account number must be generated probably get a new credit card whatever so there are a lot of steps that have to happen.

12:59A lot of banks, like other industries as well, try to automate as much as possible. They try to digitize those processes. They, of course, want to move away from pen and paper and so on and so forth. And then process automation. And I find that really important. And it took me a couple of years actually to understand that a lot of confusion comes from not understanding that. For me, process automation consists of two parts. One is what I They call process orchestration. So basically the coordination of what steps have to happen. And the other is task automation. So basically not having all the tasks done by humans and probably automating them with programming, microservices, probably calling the right API or whatever.

13:44And the combination then is process automation. And you can do it basically in two different flavors if you want to. Very often they come together, but you could do process orchestration only if you like. then that would mean that you, for example, model the process saying there are a couple of steps I have to do. And then I use a workflow orchestration engine to run that model instances of that saying, okay, the first thing I have to do is fraud check. And then if you just use orchestration, you would push that to a task list of a human saying, hey, you have to do fraud check now for this customer.

14:18Then the human does that, basically talks back to you. And then the workflow engine or the orchestration engine knows what happens next. And that's the orchestration piece. And you could do task automation only, which basically means you get an incoming new bank account opening request, and then you program a small piece of code that whatever grabs that email, grabs the data, does the front check, write something to a database. Then you probably automate the task without looking at the orchestration piece, the overall process. And process automation basically is a bit of both. And where I'm focused on a lot is the orchestrating part, basically getting order into all the chaos of things that.

14:59Yeah, and I also work on a product that is really focused on workflow automation type stuff. And I think, you know, one of the big points that I always make about this is like automation is not only a way to speed up your workflows because you're, you know, as you said, removing like that pen to paper sort of component of things, but also standardizes. So it's like you're being more consistent while also moving more efficiently. You know, so it's like there's sort of two benefits to that. I'm 100%. So we typically even put that into three or four different buckets, depending on how we talk about that.

15:31The one is the operational efficiency. That basically means you're basically can operate cheaper, probably, or also faster. Then we have the whole thing about compliance and risk management. So you basically reduce the error rate, basically, or the bias in the process and make it repeatable. Great analysis of process automation. So let's talk a little bit more, broadly speaking, about what this category itself looks like. Because depending on the challenges that you're trying to solve, the requirements you face as an organization, particularly in compliance, you're probably facing a lot of different competing philosophies around how to solve these challenges.

16:11So like walk me through like how you would break down the use cases within this category and how they fit into the different philosophies. It's actually a good segue to answer the last category of the last question because that's a lot of things are around customer experience nowadays. Like how quickly can you do things? Don't get anything lost. I think we all know that from a customer perspective. You ordered something, you never received it. My last time I opened up a bank account with an actually big German bank, it took like six weeks and I called in two times, I think. So there you get lost between certain cracks.

16:47And that's a really bad experience. And especially if you now look at how you implement those processes behind the scenes, it's sometimes understandable why that happens. I mean, the obvious thing is you don't implement or you don't look or focus on the end-to-end process at all. So that basically means it's kind of a chaos ping pong. You might do that chaos ping pong more automated. Like nowadays, I'm not sure if you heard about RPA, robotic process automation. That was quite a hype. I would say five or six years ago. So the idea was instead of a human entering data in a user interface, we use a bot doing that.

17:25So the human doesn't have to do that. But the problem with that is it looks at the task automation part. It doesn't look at the overall process. so it's still chaos underneath we had a good use case i i love that with dolce telecom and they they had a lot of human word then they introduced a massive amount of bots and then they ended up with what they called a spaghetti bot integration and then they started to understand that it's a bit like the shit in shit out idea you have to basically understand your process first then you have to know how you want to automate the process and then you can go about that so you have spaghetti bots which doesn't help you so you want to want to understand the process layer and actually there are a couple of other technologies out there which go in a different problem for me event-driven architectures is one of those technologies or let's say architectural styles very often combined with microservices where the idea was we can have all those microservices they emit events like hey somebody opened a bank account hey we cleared fraud for for certain things and then we have other microservices reacting to those events and the same problem now you start to automate the process yes you start to automate the task but you don't understand the process level anymore why does certain thing happen in a certain sequence so these are typical problem than you get if you don't have that orchestration there on a separate view.

18:53I know, and I really like how you opened this with focusing on the customer experience, thinking about how that process goes all the way to the very end user that is impacted by it. I'm very much a subscriber to Tom Peters, Thriving on Chaos, where the rapidness in which you respond to your customers, and a customer can be internal as well. It can be internal, external, whoever it is, but using automation to be, to having that full-fledged process that can just respond more quickly to customer needs. And one part of that, and you can put it in the customer experience box, but you could also put it in the business agility box.

19:28So we see a lot of change nowadays, or a change in the way the customer interacts with the organization, websites a couple of years back, now it's chatbots or whatever it is tomorrow. And we also see a lot of changes in the business model. So there is a lot of change coming in order to make that happen and to, on the one hand, make the customer still happy and, on the other hand, also keep the company in business. You have to be able to adjust and to change those processes. And the first step is always to understand that. But if you have this big bowl of spaghetti integrations or bots or AI agencies, if you're super modern or event-controlled microservices, it's super hard to understand what's going on.

20:14And if you want to change something, that's a big, big problem, actually. And that's where process orchestration starts, where you say, OK, let's extract that process. Let's understand it and let's control it on a, let's say, on a separate layer like that. extract it from the spaghetti, and then you start understanding it. I'm not so much love talking about layers because sometimes people then think the pros orchestrator is a big, huge, monolithic thing above everything else, which is not necessarily true. Yeah, the system architecture, models, stuff like that. You can still find a good architecture.

20:50You can still go with microservice, for example. You can have one microservice owning the bank account opening, for example, but then also owning that process and still use a process orchestrator and a visual model to make that visible, understandable, and changeable. So I think you've done a great job explaining this and convincing our audience that you've clearly thought about this for quite a long time. What's your unique take on this? You know, obviously you started a company, the community that is focused on this problem. So what's the unique take that you believe you've brought to the table?

21:23When I started the company, basically, we were a consulting company and we were focused around the whole, back then it was named business process management, so basically business processes. And we were working with a lot of different clients actually using different kinds of software that ranged from these, back then it was BPM suites like IBM or others, or very, let's say, low-tech, open source stacks sometimes. What we understand after a while is that there is a lot of value in these process models and executable processes, but the software is really, really a problem that we have out there.

21:59Either too monolithic, too complex, too hard to work with for a developer, or the open source ones were too low-tech, basically. They didn't have a, not even a visual editor or a visual model. Probably you had to code a lot of things around that, so not very productive. And from that experience, we said, okay, this is how we think such a process orchestrator should look like. And one of the key elements there, I mean, besides the obvious things, you have a visual workflow model, you can execute that directly. We're using BPMN as a standard, which is quite adopted worldwide. That's the obvious things.

22:38But then we made it what we call developer-friendly and very open in the architecture. So you can plug that into any kind of technology you like. I mean, they're rooted in Java. So Java is sometimes a bit easier, but we also have clients in the.NET world or nowadays Python gets more and more important. So you can hook that into the architecture used there, into the CI, CD pipelines. So basically the software engineering approach you're doing and also hooking it into whatever you have like SSO and so on and so forth. So it's very easy to integrate that into your software development approach and not the other way around.

23:15Because what we see with a lot of those orchestration vendors, they're very often much more rooted in low code, which has more the vibes of we don't want to have the developers anymore. We want to do that with the business people only, and then we just can do that magically out of the box. And that was not working. That's what we saw back then, especially not for complex processes. And we had the exact opposite approach saying, okay, there's value in all those models, but let's take it to the developer world and make it accessible there. And then probably add some layers of low code if people want to.

23:49As someone who used to be a software developer, I've encountered this low code versus software development conundrum many times. And from my experience, I think most developers have a reaction that ranges from eye rolls and groans to like outright hostility, depending on the situation. I think that's accurate, yeah. Yeah, so let's just start at like a high level. So if you're a company that is trying to figure out when to deploy low code versus actually having your software developers focus on it, on like building something custom, like at a high level, how do you break down where you draw that line?

24:27First of all, to acknowledge that I'm on the same page, actually. And when we started the company and when we moved into being a product vendor more than 10 years ago, then it was exactly that what drew us there. It shouldn't be always low-code just because you'd want to use BPMN, for example, to automate workflows. The thing that changed over time now is that it's not such a binary decision anymore. Back then, it was you were either a business person and you want to use low code or you were a developer and you want to write Java. But nowadays, we have also in the roles of the people, we have a much more diverse set of people that range from really senior developers or whatever.

25:09Super spring guru, for example, to probably very junior developers or people that are more in, for example, data models and they can do a bit of Python or so on and so forth. We have business people that also understand certain bits and pieces of coding. We have AI-assisted coding that people with a bit of technical understanding can probably use. So it opens up a big variety of people, actually. And so it's not so much about low-code or not low-code. For me, the important thing is to understand that you also have to build very different solutions in the organization. So that can range from super complex.

25:48So we typically use those kind of axes like complexity, criticality, value to the business. And in complexity, you can then find a lot of things like, for example, how often does that process run? It's a very different thing if you do something once a day or if you run it a million times per second. So we have customers doing that, which needs, of course, if it breaks, it's a different order of magnitude, the problem. or how many different systems or we like to call them endpoints are involved, different types of systems. It's a bit of just calling some SaaS services or do I have my bespoke AS400 host system and seller which I have to talk to.

26:32So all these kind of things matter. And then you get kind of a range where you say, okay, my use case is here. For example, I have a core bank account opening process that runs a couple of thousand times per day. If I get in trouble, it's really a problem. I have compliance requirements. So it's important to get it right. I want to get to a certain quality. And then you use things from software engineering because that's what we know in software engineering, how to do high-quality software. We want to write tests, automated tests, for example, and pipelines and so on and so forth. But at the same time, you will have use cases on the other end where, say, in banking, one of the examples I often use is the bank account, bank transfer limit change, which you can do in your online banking.