For the better part of a decade, the product designer's workflow followed a familiar rhythm: craft high-fidelity screens in Figma, build an interactive prototype, document the design system, then hand everything to a developer who translated pixels into working software. The handoff was the industry's ritual — and its most notorious point of failure.

Vibe coding cuts through that sequence. The practice describes developing fully functional software via natural-language prompts, on-the-go code generation and an iterative feedback loop. Rather than writing complex lines of syntax, builders explain the vibe, behaviour and logic of an interface, and AI models do the rest, generating, executing and iterating the code.

Ayomide Olukoga, a Lead Product Designer specialising in digital product strategy, interface architecture and human-centred design systems, does not see vibe coding displacing designers.

"When anyone can generate functional code in seconds, code itself becomes a commodity," he says in an interview with TechNext24. "What becomes rare and valuable is design judgement."

That distinction may determine which designers thrive as AI code generation becomes a standard production tool. Olukoga, who spent years inside the traditional design-to-development pipeline and now builds software differently, believes vibe coding represents the most consequential shift in software development since graphical design tools were invented.

A Prototype's "Well-Mannered Illusion"

A Figma prototype, Olukoga argues, is a well-mannered illusion. Anyone can animate a button hover state in a design tool, but there is no way to simulate a live database query or user validation without writing actual code. Vibe coding collapses that distance entirely, and the line between a design prototype and a live application vanishes.

A designer can describe an interaction naturally: "Build a dynamic filtering bar for a property search app where selecting 'Pet Friendly' triggers a subtle layout shift and updates the list cards in real time with smooth CSS spring animations." Within seconds, the AI returns functional React or HTML/CSS code. Designers are no longer merely presenting visual mockups of what software should look like; they are handing over living, interactive codebases of how software actually works.

"Vibe coding turns the designer into a film director," Olukoga explains. "The AI acts as the camera crew and set builders executing the vision, but the designer maintains total creative direction over the final cut. Many people use tools like Claude Code, Cursor, or Anti-gravity to create things, but we still need human judgement, which is something AI cannot provide at the moment. AI cannot make decisions based on human psychology."

The iteration loop compresses accordingly. When a transition feels sluggish, the designer does not hunt through hundreds of lines of animation code; the instruction is conversational — make the background blur darker and increase the entry spring animation by 200 milliseconds. A process that once took days now takes seconds.

When Technically Strong Code Still Fails Users

Technically sound code is not the same as a good product, Olukoga cautions. He offers a case in point: a game he built with a friend, using Claude to generate the initial prototype.

"Claude generated visually and technically strong code, but it lacked good UX for a general audience," he says. The fix required human intervention: a Nigerian illustrator was brought in to create mascot artwork that resonated with the local market. "Claude generated sound code, but it was visually lacking in cultural branding without our research and human input."

Asked for a recent example of this directing role in action, Oluk points to a live engagement in the Middle East, where he works as a product designer for a telecommunications company tasked with redesigning a website with a very large architecture.

"I am currently working as a product designer for a telecommunications agency in the Middle East. I was tasked with identifying the ideal entry point for a redesigned website, which features a very large architecture. I asked Claude to crawl the current website, map out the site architecture, and identify the flawed entry flows. It brought out multiple entry points, and based on user interviews, we deduced which ones would convert the best. Because user interaction in the Middle East differs from the UK, we factored that in, and Claude helped identify new entry points that allowed us to make informed decisions moving forward," he recounts.

It is a pattern he sees repeating across the industry: the AI maps terrain at machine speed; the human decides which direction to walk.

That boundary is important as prototypes become executable. Faster generation does not remove the need to define the right problem, evaluate whether flows work for intended users, or account for cultural context. In Olukoga's examples, those decisions come from user interviews, research and product judgement rather than code generation alone.

Displacement Lands on Tasks, Not People

Anxiety runs high in design circles, where the concern is that as AI tools make coding accessible to everyone, the value of dedicated product designers will diminish. A recent statement by IMF Managing Director Kristalina Georgieva noted that AI would displace as much as 40% of the global workforce.

Olukoga argues the reality is the opposite. "In reality, the exact opposite is happening. When anyone can generate functional code in seconds, code itself becomes a commodity. What becomes rare and valuable is design judgement," he says.

"AI cannot feel," he adds, recalling a discussion with designers last year. "I tell them that right now is the worst AI will ever be, as it will continually evolve."

The displacement, in his reading, lands on tasks rather than people. "Rather than replacing 40% of designers, AI is directly liquidating 40% of standard, routine, execution-heavy tasks. AI cannot replace problem framing and system architecture. It takes over heavy tasks, producing results in minutes, and the designer simply uses their higher taste to iterate and make decisions."

An AI can generate ten variations of a checkout screen instantly, but it does not intuitively understand human psychology, spatial harmony, accessibility or brand nuance. "AI will replace those who refuse to use it in their workflow," he argues. "AI doesn't know why a customer abandons a cart on step three; it takes a skilled product designer to curate, critique, and direct the AI toward a seamless user experience."

What Happens to Figma

If AI can generate production-ready interfaces from prompts, what becomes of Figma, the default tool of product teams everywhere?

"I don't see vibe coding entirely replacing Figma, but it fundamentally changes what Figma is used for," Olukoga says. In his view, natural-language vibe coding eliminates large chunks of high-fidelity screen production, static prototyping and developer handoffs. It will replace Figma for mechanical production execution, but not for strategic alignment or systemic design. Designers once spent 70% of their time drawing static screens and exporting specs; vibe coding absorbs that mechanical production, making it easier for designers to focus on decision-making while AI handles UI frame generation.

New Economics for Early-Stage Teams

For early-stage startups in emerging markets, where engineering talent is expensive, vibe coding alters the economics of team-building. Olukoga describes a structural shift: the delayed hiring of a dedicated front-end engineer. Historically, a startup needed a backend architect and a frontend engineer just to get mobile components built. Natural-language generation trivialises boilerplate code, meaning a single full-stack founder can now build and deploy complete user interfaces. Founders can ship an MVP quickly and hire technical staff only when scaling demands it.

"Vibe coding is bridging the historical divide between design and engineering," he notes. "It allows designers to move directly from idea to working software without getting bogged down in syntax errors, boilerplate code, or lost-in-translation handoffs."

Learn Both Tools, Keep the Judgement

For designers entering the field today, Olukoga's counsel is not to choose between old tools and new ones. "They should learn both," he says. "Because AI generates so many codes and designs, human judgement is required to yield unique results." Mastering complex Figma components builds design fundamentals; prompting AI tools accelerates execution. The judgement, though, is non-negotiable.

The design profession's centre of gravity is moving away from static pixels and toward something closer to direction. Software development, Olukoga concludes, is no longer about drawing what an interface should look like, but about directing the vibe through human judgement. The designers who thrive, he argues, will be those who pair deep empathy for users with the ability to prompt, guide and shape live code in real time.