In our latest research, 76% of tech leaders believe AI will make low-code tools more efficient, and that single statistic captures a bigger shift underway: AI isn’t just accelerating software delivery, it’s reshaping what it means to be a “software engineer.”
In this episode, I spoke with Ray Kok, CEO of Mendix, about how agentic AI is colliding with low-code, why traditional programming skills are becoming less central, and how enterprises can scale development without turning low-code into shadow IT. We also unpacked the rise of fusion teams, the increasing importance of non-functional requirements (NFRs) as a differentiator, and why the developers who thrive next will look a lot more like today’s software architects.
Why the Software Engineering Skill Stack Is Being Disrupted
Ray’s premise is direct: software engineering is undergoing fundamental change because AI is becoming “pretty good” at the mechanics of producing code.
“AI is actually changing the game of software engineering,” he told me. And this disruption is happening on two fronts at once. On the pro-code side, we’re seeing agentic IDEs automate tasks developers traditionally handled manually. On the low-code side, the same pattern is emerging—only it starts from a model-based foundation.
“In Mendix… we’re turning our IDE experience into an agentic IDE experience,” Ray said, “where AI is actually driving major steps within the software development lifecycle.”
This forces a hard question: if AI can generate functional code reliably enough, what skills remain the durable foundation of a software career?
Ray’s view is that traditional programming fundamentals won’t disappear, but they will become less differentiating. The center of gravity shifts away from writing code line-by-line and toward orchestrating AI systems and validating what they produce.
“We’re Not Makers Anymore—We’re Orchestrators”
One of the most useful frames Ray offered is that knowledge work in software engineering is changing shape.
“What used to be the knowledge work of today is not going to be the same as the knowledge work of the future,” he said. “We’re not going to be makers—we’re going to be orchestrators.”
In this model, developers spend less time on the raw output and more time ensuring the output meets enterprise expectations, especially non-functional requirements like security, scalability, performance, reliability, and compliance.
“People got measured on how good of a programmer you are,” Ray added. “Now it’s going to be: how good of an orchestrator are you… and how well can you build the non-functional requirements into whatever you’re developing.”
That’s a meaningful shift in career paths and hiring patterns: the best future engineers won’t be the fastest coders; they’ll be the strongest system thinkers.
Low-Code Doesn’t Win Unless It’s Enterprise-Grade
Ray also drew a line between consumer-style “no-code” and enterprise low-code. With the rise of vibe coding tools, he argues that classic no-code platforms face an existential squeeze because anyone can now generate “something that works.”
“Anybody can actually build some meaningful application,” he said. “Therefore, the category of no-codes… they have a very short run.”
But the enterprise doesn’t just need “something that works.” It needs production-grade software that satisfies NFRs and can be governed at scale.
“That foundation… is not about functionality,” Ray said. “It’s about the non-functional requirements.”
From the Mendix perspective, the opportunity is to marry agentic development experiences with enterprise-grade foundations so teams can accelerate delivery without losing control.
Fusion Teams Are Evolving, and So Is the Professional Developer Role
In our research, we’ve tracked the rise of fusion teams as a practical response to skyrocketing demand for applications. Ray describes fusion teams as a blend of business subject-matter experts and technical specialists working together using low-code as a shared delivery platform.
The key shift now is who the “technical specialist” is.
“Instead of… lower level engineers, we’re now talking about the architects working with the business,” Ray said. And as AI accelerates the SDLC, business stakeholders can take on larger parts of delivery, while the technical role becomes more about guardrails, composition, reuse, and governance.
To Ray, the result is a different operating model: better alignment between business and IT, faster iteration, and a more scalable approach to enterprise modernization, particularly as organizations try to meet the explosion of demand (including edge application growth and internal automation backlogs).
Ask the Right Questions, Not Just Give the Right Answers
Ray made an important point about certifications and traditional skills screening: advanced LLMs can already “beat” many standard certification exams. That doesn’t mean expertise is obsolete, but it does change what hiring should optimize for.
“Instead of giving all of the right answers,” he said, “what is now much more important is… people that can actually ask all the right questions.”
In an AI-driven SDLC, the ability to shape outcomes depends on framing: translating functional requirements and NFRs into the prompts, constraints, and workflows that guide AI systems toward production-grade output.
Ray’s implied hiring criteria shift looks like this:
- Systems thinking over syntax mastery
- Architectural reasoning over code volume
- Ability to define and test NFRs
- Comfort operating at higher levels of abstraction
- Skill at orchestrating tools/agents and validating outcomes
In his view, the “software engineers of the future” are closer to what we call software architects today, focused on composition, reuse, and production fitness.
Analyst Take
This conversation is a clear signal that the next era of software engineering will be defined less by “who can write code fastest” and more by “who can deliver production-grade systems with AI in the loop.”
Three themes stood out:
- AI will compress coding as a differentiator.
Coding won’t disappear, but it becomes a lower-level instruction layer—not the main unit of value. The differentiator becomes orchestration and validation. - Non-functional requirements become the competitive edge.
Security, performance, reliability, scalability, and governance are the hard parts—and they’ll define whether AI-accelerated software is enterprise-ready or just prototype-grade. - Fusion teams scale delivery, but only with guardrails.
Empowering business users is inevitable. The winning model is not “shadow IT with copilots,” but business + architecture-grade technical leadership on a platform designed for production.
My guidance for practitioners: invest in architecture skills, NFR thinking, and AI workflow orchestration. For IT leaders: update hiring and enablement around systems reasoning, governance, and higher-level abstraction skills, not just programming ability.
