Gordon Van Huizen, SVP of Strategy at Mendix – Interview Series

Gordon Van Huizen is SVP of Strategy at leading low-code provider, Mendix, a Siemens business. In this role, Van Huizen identifies and explores strategies for emergent technology advancements, and works to incubate product innovations within Mendix, all with a focus on how these technologies can impact and bring value to customers.

How is AI permanently altering the software development lifecycle (SDLC), especially in low-code and no-code environments?

AI is really shaking up the software development lifecycle, especially as we leverage natural language more and more. Instead of writing lines of code, organizations are starting to define and build software just by describing expectations. It’s becoming more about expressing intent and having a conversation with smart tools that can take that intent and turn it into code, interfaces, and even tests.

As AI continues to weave its way into the SDLC, I think we’ll start to see how powerful that shift really is. Communicating what we want rather than how to build it will feel more natural, and honestly, be more lasting than writing code the traditional way. Eventually, code as we know it might fade into the background. And not just that, we’re heading toward a whole new model of software that’s not just built by AI, but is intelligent in its own right. The shift is big and exciting, and perhaps the most significant shift in software we’ll see in our lives.

What role do you see Agentic AI playing in future application development, and how should developers and platform architects prepare for its observability challenges?

As Agentic AI keeps reimagining the SDLC, we’re not just seeing faster, cheaper, and higher-quality development, we’re also seeing that development become more accessible. People can get creative and experiment without needing to be expert coders; all they need to be able to do is clearly express what they want. Still, all that power comes with complexity. The software we’re building today is more advanced than ever, which brings new challenges, especially as multi-agent platforms continue to emerge. Interoperability is becoming a headache because applications are naturally distributed and often involve tools from other vendors and tech stacks. That’s where low-code platforms really start to shine.

They can automate a lot of grunt work on the deployment side while giving you a clear view across the whole system. Then, once you have that observability layer in place, you can bring AI into the mix to help make sense of what’s going on. AI can surface issues like performance drops or inaccurate outputs and explain the root cause in plain language. That kind of clarity is a game-changer for both developers and operations teams. All of this means we’ll need low-code more than ever, as its very nature addresses these challenges head-on. In particular, we’ll see the powerful combination of AI-augmented development and low-code. You can express yourself in natural language, then see the results in a visual way — including data, logic, and user interfaces — and interact through any combination of natural language and the visual IDE to further refine and expand upon the generated software.

Do you believe the traditional concept of “developer” is evolving due to low-code and AI? What skills will be most critical in the next decade?

Today, software developers and AI engineers are often seen as two separate roles, but we’re already starting to see some overlap, both through developers learning the skills required for AI engineering and with fusion teams that bring together developers, AI engineers, data engineers, and even data scientists. Honestly, that kind of collaboration is exactly what we need right now. But, yes, the traditional concept of “developer” is certainly evolving. It’s only a matter of time before software developers become AI engineers. Ultimately, AI engineering is still software engineering; it just involves a set of tools and concepts that many developers haven’t worked with yet. Those skills are learnable, and many traditional developers will likely find this new work exciting. It opens the door to building smarter, more dynamic solutions, and that’s a rewarding direction to grow in.

How does Mendix balance the accessibility of low-code with the complexity of building AI-powered applications?

Mendix’s goal is to alleviate the complexity of building AI-powered applications while also ensuring that what developers build today is future-proof. We want to make things simpler without removing the flexibility developers need. We leverage a visual approach so you can actually see how the agents and systems fit together, like if one agent triggers another. With Mendix’s low-code tools, the architecture and behavior of these AI-infused systems are laid out in a way that doesn’t feel like a complex multi-agent system. It just looks like a clean, understandable application.

How are low-code platforms like Mendix enabling non-developers to build sophisticated AI-driven solutions, and what are some of the best examples you’ve seen?

At Mendix, we meet developers, business technologists, and citizen developers where they are with respect to their understanding and needs for AI-infused apps; the platform’s tools are easy to pick up and use at the onset. We guide them through the experience step by step until they’re using low-code to build smart, AI-powered applications that are every bit as advanced as those built with high code. They start by building prompts using our low-code prompt builder. Once they’re comfortable with that, they can ground their generative AI-infused app with data specific to the business or solution with a built-in low-code knowledge base. And when they’re ready for it, they can even build AI agents through low-code orchestration and tool use.

One of the best real-world examples is the AI-native global payroll platform built on Mendix, datascalehr. Payroll, especially as it varies from country to country, is notoriously complex, with constantly changing regulations, compliance requirements, and vast amounts of data. Using Mendix, datascalehr founders rapidly developed a next-generation platform that leverages AI for intelligent automation, compliance checks, and contextual assistance. What’s powerful here is that business technologists and domain experts — not just professional developers — were able to shape how AI features were embedded, ensuring the solution directly addressed customer needs. Low-code is making sophisticated, AI-driven solutions both accessible and enterprise-ready.

Can you walk us through how AI is being used within Mendix itself — both in how the platform is built and how it empowers users?

“Create with Maia” is Mendix’s answer to both infusing AI into the application development process and enabling our customers and partners to build intelligent, AI-powered applications. Recently launched with the latest version of Mendix, Mendix 11, Maia allows users to easily create, orchestrate, and deploy AI agents and multi-agent applications throughout the entire software development lifecycle. Even before users start building, they can leverage Maia and use natural language to ensure goals, success criteria, and user stories are aligned before creation. Create with Maia also helps turn brainstorms, mockups, diagrams, and requirements into clear, actionable project plans. Then, once the initial software is created, users can quickly refine that software with the inherent speed of low-code. The result is fewer iterations, faster delivery, stronger governance, and software that’s built right from the start.

How are you seeing AI and low-code come together to support nonprofits or mission-driven organizations working to solve social or environmental issues?

AI and low-code are incredible tools for tackling real-world challenges, mostly because they give personnel focused on solving critical social problems an option to innovate, even with limited budgets and technical skill levels. One example that really sticks out to me is from Alliance for Orphans (A4O), a nonprofit based in San Antonio that offers respite babysitters for foster families. The company hit a big roadblock when they realized it was tough to find, train, and certify babysitters, essential for helping foster parents get the support they need. Low-code helped them build an application to streamline the certification process, bringing together systems across different agencies, digitizing paperwork, and building a centralized database to track certified babysitters. The application helped A4O certify 81 respite babysitters, and since its inception, the applications have only continued to grow. It’s such a powerful example of how low-code can make a real, positive difference in people’s lives, and that was just one example.

What are the unique challenges and opportunities of using synthetic data within a low-code environment?

Synthetic data inherently reduces privacy risks since it does not contain real personal information, making it easier to comply with data protection regulations (such as GDPR) and minimize legal exposure. Of course, utilizing synthetic data is also faster, cheaper, and easier than building data sets from scratch and labelling the data for use by AI, which may be out of scope or impractical for some projects.

That said, synthetic data may contain inaccuracies, bias and toxicity, and even fail to capture the noise, outliers, and full range of scenarios inherent to real-world use — leading to potential failures in production. Therefore, it’s necessary to put guardrails in place and establish a rigorous testing and validation approach, one that extends the application testing process to include validating AI output. For business-critical systems, it’s also important to keep humans in the loop so that they can apply their own discernment, optimally providing feedback from the application itself.

How do you see the convergence of IT and OT evolving when AI and low-code tools are introduced into operational settings?

The power and accuracy of any agentic AI solution boil down to context; the quality and volume of data are critical. That’s why it’s becoming essential for those in manufacturing, energy, and other industrial segments to have a solid data foundation that brings together both IT and OT data. Unfortunately, OT data isn’t always easy to work with. For example, it’s often unlabeled with no clear metadata or schema to guide you. The good news is that specialized tools are available for transforming OT data and augmenting it with the necessary metadata, preparing it for use within intelligent applications via AI-augmented creation of suitable data models. Once ingested, the OT data can be used alongside and combined with IT data for use within applications and to provide context to generative AI.

As a former Gartner analyst and now SVP of Strategy at Mendix, how do you separate AI hype from truly transformative innovation when shaping your product roadmap?

Separating AI hype from true innovation requires a disciplined and pragmatic approach, but it’s a procedure I’ve fine-tuned as trends come and go. First and foremost, I engage directly with customers and prospects to understand their real-world plans and requirements — that is, what they actually need to move their business forward. The Mendix product team also takes a test-and-learn approach by delivering MVPs of new capabilities and then working closely with customers to gather feedback and validate whether these innovations actually deliver tangible value. As you can see, collaboration is a key way of sorting through AI hype, so I also like to actively work with our existing partners and explore potential new ones to bring in additional perspectives and expertise.

Finally, I draw on my experience with current and previous waves of emerging technology, keeping a close eye on maturity levels and adoption curves. This really helps filter out what’s speculative versus what’s likely to gain traction, so we can prioritize investments that will drive long-term impact for our customers.

Thank you for the great inteview, readers who wish to learn more should visit Mendix.