Why do we actually care about AI Frontend Dev right now?
With AI becoming part of almost every tool we use — Figma, Canva, Adobe, and beyond — the urgency to understand where it adds real value (and where it doesn’t) has never been greater.
And instead of leaving it at theory, we did what engineers do best: we tested it.
Two of our engineers, Anton Herasymenko and Ivan Datsunov, set up an internal experiment. The goal was clear: assess how far AI can go in converting static designs into functioning code without much manual input.
The setup was intentionally simple. We took a set of Figma screens — representing the client’s idea for a new application — and gave them to a coding assistant. No database, no APIs, no fully mapped-out logic. Just the design and the challenge: Can we generate meaningful front-end code with minimal developer effort using AI tools?
The intention wasn’t to build a production-ready app. Instead, we wanted to understand the boundaries. Where does AI help? Where does it fail? And more importantly — how can we use this insight to work smarter?
IIt didn’t take long to see where AI tools like Cursor and similar developer assistants could genuinely accelerate progress.
The models handled basic layout structure well. Things like forms, buttons, standard input flows, and container components were generated with a fair degree of accuracy. In terms of raw speed, this was impressive. What would normally take half a day of repetitive coding now took a few minutes.
This is where AI frontend dev shows real promise — accelerating routine UI generation and giving teams a faster starting point. More importantly, these auto-generated components served as a working base for further development. Developers could quickly modify or refine what was generated instead of starting from scratch. That kind of head start is valuable in MVP stages or fast iterations.
As Ivan Datsunov explained during the debrief, “If it’s a straightforward UI, AI can absolutely help you move faster. It won’t get it perfect, but it’ll get you 70–80% of the way. And that’s already a win.”
The moment we moved beyond the visual layer, things became more complicated.
Any time the application needed to reflect non-standard behavior, custom workflows, or domain-specific logic, the AI struggled. It lacked the context to interpret what the business actually needed. And the effort it took to prompt and re-prompt the model into doing the right thing quickly outweighed the benefit of using it at all.
As Ivan pointed out, "You start spending more time trying to explain what you want than if you just wrote the code manually."
This aligned perfectly with our corporate statement: AI doesn’t magically increase productivity just because it’s there. It only helps if you’re solving the right problem in the right way.
This experiment wasn’t just about exploring a cool feature. It was a direct response to an open question within our team. It was also a proof of how we approach new technologies at Opinov8, not with blind enthusiasm, but with curiosity, structure, and a focus on real outcomes.
Christian Aaen, our co-founder, framed it well: “AI won’t solve everything. But if we learn where it fits, it will absolutely change the way we work.”
He also raised an idea that might shape our next experiment: Can we improve AI output by training it on our existing codebase? If a model could learn our architecture and coding conventions, would its suggestions become more useful? That’s a conversation we plan to continue.
We’re now preparing a similar experiment on the backend side — where logic, performance, and structure are even more critical. Following our AI Frontend Dev test, where we focused on generating UI components, the next step is to see how AI performs when it comes to state management, authentication, API handling, and more.
This is part of a larger, ongoing effort across Opinov8 to understand how AI should fit into our delivery pipeline — at every stage, from discovery and design to deployment and scale.
And as always, we believe the best insights come not from reading whitepapers, but from building things ourselves.
AI Frontend Dev refers to the use of AI-powered tools to automate parts of frontend development — like generating layout, components, or boilerplate code from design files.
AI tools can generate basic UI components and structure from Figma designs. However, they often fall short when it comes to complex logic, data flow, or domain-specific requirements.
No, the goal wasn’t to launch a production-ready app. The experiment focused on understanding where AI adds value in the early stages of frontend development.
AI tools are helpful for speeding up repetitive tasks and scaffolding UIs. But they struggle with custom logic and need close developer supervision.
Opinov8 have been named the Best Software Development Agency in Europe at the prestigious Netty Awards. The Netty Awards honor top innovators in the digital world, showcasing the best in technical expertise, creativity, and groundbreaking solutions. This recognition underscores Opinov8’s role as a leading force in the software development space, helping businesses across Europe transform […]
Opinov8 have been named the Best Software Development Agency in Europe at the prestigious Netty Awards. The Netty Awards honor top innovators in the digital world, showcasing the best in technical expertise, creativity, and groundbreaking solutions. This recognition underscores Opinov8’s role as a leading force in the software development space, helping businesses across Europe transform […]
RPA (robotic process automation) tools are great solutions for automating repetitive workflows. Look up how integrate RPA into your company’s existing set of tools.
RPA (robotic process automation) tools are great solutions for automating repetitive workflows. Look up how integrate RPA into your company’s existing set of tools.
