By focusing on the skills that large language models lack, ‘designgineers’ can adapt to a market upended by AI.
“The money [in front-end web development] is … gone.” This, according to Santiago Valdarrama, a leading voice on AI and software development. Not everyone agrees, of course—just look at the online storm he kicked off recently on X. Still, Valdarrama has a point. “Two years ago, everyone was throwing money at you to build websites,” he tells his developer audience. “Those clients are now gone.”
Where have they gone?
To large language models (LLMs), of course. Or, more truthfully, to front-end developers who know how to build with LLMs. This shift is disproportionately benefiting more senior developers, as I’ve noted, but there’s potential for it to benefit every front-end or full-stack developer. The trick is not to panic. Adapt.
LLM barbarians at the gate
It’s hard to overstate how fast AI coding tools have advanced. “Every week, we get a new [model that is] leaps and bounds around everything we’ve seen,” argues developer Valdarrama, marveling that LLMs are now solving “problems that nobody—and I mean nobody—thought possible a few years ago.” He continues, “Large language models have blown past every limitation I had imagined,” to the point where “today, you’d have to be crazy to hire untrained people to write code” because the bar to contribute has been raised so high by AI.
When even seasoned engineers are surprised by how far AI can go, it’s easy to see why junior developers fear they’ll be automated out of a job.
Indeed, modern LLMs can churn out front-end code on command. Vercel CEO Guillermo Rauch notes that by the time GPT-3 arrived, it was “very good at being able to design with HTML and CSS and also [at] writing React code,” hinting at a revolutionized web development process. That promise has only grown with GPT-4 and other more advanced models.
But that doesn’t mean LLMs eliminate the need for front-end developers. At least, not for experienced ones.
Despite impressive demos, LLMs today are far from a drop-in replacement for skilled front-end engineers. The flashy examples usually involve generating a small application or a static page layout. In real production apps, front-end work is deeply entwined with context, and context is where current AI falls down. “The variation between each team and project, … [the] combination of tools, frameworks, libraries, [and] CSS … are infinite,” Ofer LaOr explains. An LLM has “zero context” about your project’s specific stack and style guidelines. In other words, an AI might produce a generic <Modal> component, but integrating it into your app’s unique architecture is still a human task.
Even at the coding level, LLMs have well-documented accuracy issues. They often generate code that looks confident but contains subtle bugs or omissions. “LLMs aren’t able to validate their assumptions or test their hypotheses. They can’t confirm whether what they’re saying is true,” observes front-end educator Josh W. Comeau. Unlike a human developer, the AI doesn’t truly understand the code’s behavior or the UI/UX implications; it’s predicting likely token sequences. The result is that “there will always be some level of inaccuracy” in AI output, Comeau concludes.
For a non-programmer trying to build an app entirely with AI, these hidden errors are land mines. “If you’re not a programmer, you won’t be able to tell which parts are accurate or not,” Comeau notes. Hence, “to use AI effectively in software development, you need sufficient experience to know when you’re getting garbage from the LLM,” as I’ve argued. Generating a 50-line demo is one thing; building and maintaining a complex web application with tens of thousands of lines of code is something else entirely.
Perhaps the LLMs will get there, but they’re not there yet.
More problem solving, less duct tape
All this points to a new understanding of the front-end developer’s role. If cranking out basic UI code becomes commoditized, the real value of a developer lies in high-level problem-solving, creativity, and polish. “Coding is how we communicate our ideas to a computer, but that has never been the hardest part,” Valdarrama notes. The hardest parts are deciding what to build, why to build it, and how to design it elegantly. The most competent engineers earn that title “because they consistently do three things better than everyone: 1) identify the right problem to solve, 2) frame the problem so the solution becomes inevitable, 3) shape the solution into an elegant, maintainable design.”
Those are precisely the human strengths that AIs, which lack true understanding or intuition, cannot replicate.
Front-end development has always involved a lot of “glue” work: integrating APIs, tweaking designs, resolving ambiguity in requirements. These are “very human problems, [which] require human operators to drive them,” as LaOr emphasizes. For instance, a typical senior front-end dev might take an “impossible” Figma design from a designer, then negotiate and iterate on it until it fits reality—a delicate dance between technical constraints and user experience that no AI currently handles. In the future, junior devs (or even non-devs) might ask AI to generate UI components, but senior developers will coach and correct the AI, weaving those pieces into a coherent, performant application.
What to do about AI
So, how should front-end developers meet this changing market?
First, embrace the AI tools rather than avoid them. As Valdarrama advises, “learn as much as possible about AI and how to use it to write better code.” The developers who pair their JavaScript or mobile framework skills with prompt-engineering savvy will outpace those who stick strictly to manual coding. In practical terms, this could mean using Copilot during development, integrating an LLM into your editor for quick fixes or leveraging AI to generate initial UI drafts that you then refine.
Second, focus on the fundamentals that AI can’t automate. This includes your understanding of user experience design, accessibility, performance tuning, and architectural principles. AI might generate a navigation menu for you, but you decide if that navigation delivers a good UX or if it meets your site’s performance budget. Strengthening your design collaboration skills is also key. Front-end devs who can speak the language of designers (and even use AI design tools) will be highly valuable in bridging the gap between vision and reality. We’re already seeing front-end development job postings evolve from “front-end developer” to “front-end architect.” Embrace this shift.
Third, develop your “critical eye” for AI output. Treat every AI suggestion as a starting point, not an absolute. If you’re a newer developer, this means you can’t skip learning proper coding practices, testing, or debugging. In fact, junior devs may need to learn more quickly now: The mundane tasks that used to be a safe training ground (like slicing a PSD into HTML or wiring up a simple form) might be handled by AI, so you’ll be challenged sooner with reviewing and improving AI-generated code. Seek mentorship and code reviews to accelerate your ability to spot mistakes. Take on those gritty tasks of validating and polishing AI code; that’s where you’ll sharpen the skills that AI lacks.
Finally, keep an open mind about the direction of front-end development. The field is likely to become more interdisciplinary. There’s talk of the rise of the “designgineer,” a hybrid designer-engineer empowered by AI tools. We see non-traditional developers creating websites via voice or visual tools. This democratization of front-end creation doesn’t mean your expertise is worth less, it means you might be guiding a larger crowd of contributors (including AIs) in bringing ideas to life. Your role could shift more toward curating and integrating components, writing high-level logic, and ensuring quality. The market for front-end developers isn’t shrinking, but the definition of the job is broadening. A globally distributed team of developers can now collaborate with the help of AI translators, copilots, and testing tools.
In effect, the world’s pool of developers is growing, and as an individual, you have more opportunity than ever to build impactful things—if you stay adaptable.
