Design Engineering in the Age of AI: Why Quality Still Matters
edited by Entrepreneur UK
|
May 10, 2026
Jakub Krehel
Artificial intelligence is often framed as a tool that speeds everything up. It helps generate, automate, and accelerate. In design engineering, however, the work has traditionally been defined by something else. Taste, iteration, and attention to detail.
These are not areas where AI naturally excels, at least not yet.
This tension is becoming more visible as AI tools enter everyday workflows. At Interfere, a company working on what it describes as the self-healing layer of the internet, the challenge is not just to build quickly, but to build systems that can monitor, fix, and improve themselves over time.
Among the team is Jakub Krehel, a founding design engineer whose work sits at the intersection of design and engineering. His path into the field started early, experimenting with visual design and development before moving into professional work across both disciplines. Before joining Interfere, he worked at OpenSea, contributing to products used by millions of people and helping build a rewards system that generated billions of dollars in trading volume.
Alongside his product work, Krehel has built a substantial following through writing and sharing design engineering insights. That combination of hands-on experience and public exploration informs how he approaches the role of AI in modern workflows.
Across both high-scale consumer platforms and early-stage infrastructure, one pattern has remained consistent.
Quality takes time.
Iteration as the Core of the Work
Building high-quality interfaces is rarely a linear process. It is shaped through iteration, where small adjustments accumulate until something starts to feel right.
That process has traditionally been slow. Much of the time is not spent building, but refining.
AI changes the pace of that loop. Instead of committing hours to a single direction, multiple approaches can be tested in a fraction of the time. Ideas that don’t work are discarded quickly, making exploration less costly and easier to navigate.
At the same time, the role of the tool remains limited.
“AI helps me explore more than ever before,” Krehel explains, “but I’m careful not to let it replace my thinking.”
AI as an Execution Layer
Where AI has had the most impact is execution.
It is effective at generating initial versions, handling repetitive work and identifying inconsistencies across large systems. Tasks that would previously take hours can often be reduced to shorter cycles of iteration and review.
In one recent example, Krehel used AI to consolidate repeated patterns across a codebase, simplifying the system and reducing duplication. The task itself became significantly faster, reducing what would have taken hours to a matter of minutes and allowing him to focus on evaluating the result rather than implementing it.
This distinction shapes how these tools are used.
“AI speeds up execution,” he says, “but it doesn’t improve judgment.”
Building Under Pressure
At Interfere, this approach is tested under real constraints.
As an early-stage company, the need to move quickly exists alongside a high bar for quality that cannot be compromised. The expectation is not just to build fast, but to build work that holds up under real use.
For Krehel, this creates a specific way of working. AI is used to accelerate early stages of development, but the majority of the effort still goes into refinement. Interfaces are adjusted, details are revisited, and decisions are evaluated against a consistent standard.
The nature of the product reinforces this. Building systems that monitor and repair themselves requires interfaces that are clear, predictable, and trustworthy. As the underlying complexity increases, the margin for error decreases.
In that context, using AI as a shortcut quickly breaks down. It can speed up parts of the process, but it cannot replace the discipline required to reach a high-quality outcome.
The Risk of Faster Output
As AI tools continue to improve, a broader challenge is emerging.
While it is now easier to produce working software, maintaining quality has become more difficult. The volume of output increases, but the overall standard does not necessarily follow.
In design engineering, quality is rarely the result of a single decision. It emerges from many small choices made consistently over time. Typography, spacing, motion, and structure all contribute to whether something feels great to use.
These decisions still require judgment.
Without that layer, the work may function, but it loses the sense of care behind it.
A Changing Role
Rather than replacing design engineers, AI is reshaping the role itself.
In Krehel’s work, the focus is already shifting. Less time is spent on producing, and more on directing and evaluating. Building is faster, but deciding what is worth building and what meets a high standard remains the harder problem.
As the barrier to building continues to fall, this shift becomes more pronounced. More people can create functional interfaces, but fewer can create ones that feel resolved.
Looking Ahead
Across industries, people tend to value products that are well made, whether in cars, fashion or architecture. Quality is often what separates something that works from something that lasts.
The same pattern is emerging in software.
As AI continues to accelerate development, the ability to move quickly is becoming less of a differentiator. What stands out is the ability to maintain quality as speed increases.
Many companies are already struggling with this balance. Output is growing, but standards are harder to maintain. Without strong judgment, faster workflows often lead to weaker results.
The tools are changing.
The standard is not.
Artificial intelligence is often framed as a tool that speeds everything up. It helps generate, automate, and accelerate. In design engineering, however, the work has traditionally been defined by something else. Taste, iteration, and attention to detail.
These are not areas where AI naturally excels, at least not yet.
This tension is becoming more visible as AI tools enter everyday workflows. At Interfere, a company working on what it describes as the self-healing layer of the internet, the challenge is not just to build quickly, but to build systems that can monitor, fix, and improve themselves over time.
