AI in Figma: an accelerator, not an autopilot
Figma has made artificial intelligence one of the main directions of its product development. The goal is not to replace experience design, but to cut the time spent on mechanical tasks: exploring a direction, generating a first structure, rewording a text, adapting a visual or producing a prototype faster.
For web projects, the change is concrete: teams can move faster from idea to testable mock-up. But that speed also creates a risk: confusing a visually correct interface with an experience genuinely suited to users, business constraints and conversion goals.
What the 2025 and 2026 studies show
Adoption of AI by professional designers is now massive, and figures published by Figma and Boston Consulting Group give a precise measure of its real impact.
Designers who say they use generative AI daily: 72%.
Among them, those who increased their use over the past year: 98%.
Designers who feel AI helps them improve their work: 91%.
Designers who feel AI helps them work faster: 89%.
Speed gain for teams using AI tools: up to 67%.
Designers who trust AI output without checking it: only 32%.
Large companies using AI in UI/UX (BCG 2025): 63%.
The most telling figure is the one about trust: despite massive adoption (72%) and a real gain in speed, only 32% of designers trust AI output without checking it. This confirms that AI changes the speed of execution, not the need for human oversight.
What the time saved is really for
The measured time saving (up to 67% on some execution tasks) does not automatically translate into faster delivery for the end client: in a well-organised agency, the freed-up time is usually reinvested in the steps AI cannot speed up, such as preparatory discussions with the client, user testing or refining accessibility. A team that uses AI only to produce more mock-ups faster, without reinvesting that time in quality, gets a faster result but not necessarily a better one: that is precisely the gap reflected in the 32% trust figure mentioned above.
The most useful AI features in Figma
Getting started faster
First-draft generation features can help you get past the blank page: a wireframe, a home screen, a simple journey or a visual variation. This is useful during exploration, when the team has to compare several directions before choosing.
Turning an idea into a prototype
Figma Make and related tools bring mock-up, prototype and exploratory code closer together. They make it possible to show an interaction earlier, test the logic of a screen, or prepare a discussion with developers.
Improving interface copy
AI can help reword microcopy, clarify a button, shorten a sentence or adapt a tone. On an online shop or a service page, these small texts play a direct role in understanding and conversion.
Retouching and producing visuals
Generation, background removal, image extension or quick adjustments can smooth creative work. Here again, the result must stay consistent with the brand identity and with the usage rights of the media.
What AI does not replace
AI suggests, but it does not naturally know your users, your legal constraints, your commercial positioning or your business priorities. An automatically generated mock-up can look appealing while forgetting an important use case, a keyboard journey, an accessible label or an essential piece of reassurance. That is exactly what the 32% trust figure reveals: designers themselves know that human review remains indispensable.
The designer's role therefore becomes even more strategic: framing the right problem, choosing the right hypotheses, checking accessibility, weighing trade-offs, documenting decisions and making sure the interface truly serves the site's goal.
What it changes for your web projects
Used well, AI in Figma can shorten the first iterations and make client discussions more concrete. It makes it possible to test more options without adding production days, with measured speed gains of up to 67% on some tasks. But a professional result still requires a method: a clear brief, priority journeys, a design system, accessibility rules, human review and validation on real content.
Visual consistency remains a matter of a controlled brand identity and of copy written to convert, which automation cannot invent on its own. AI speeds up execution; it replaces neither an understanding of the need, nor experience, nor editorial responsibility.
A concrete example: from brief to testable mock-up
On a project to redesign a service page, AI-assisted preparation usually runs as follows: an initial brief written with the client (goal, audience, constraints), generation of two or three structural options to open the discussion, human selection of the most relevant direction, then complete manual refinement (information hierarchy, accessibility, consistency with the design system, final copy approved by a writer). AI therefore comes in mainly upstream, to explore faster, never at the end of the chain on the version delivered to the client.
Three checks before presenting a mock-up
How do you know an AI-assisted mock-up is ready to show a client? The answer comes down to three checks, to be run systematically before any presentation. The first is consistency with the existing design system: colours, typefaces, components and spacing must follow the rules already set, otherwise the mock-up introduces a visual break the client will notice, even without being able to name it. The second is accessibility: sufficient contrast, possible keyboard navigation, a logical heading structure, explicit labels. Automatic generation almost always ignores these requirements, which are a legal obligation as much as good practice.
The third check, the one most often neglected, concerns the content displayed. A generated mock-up is full of filler text, fake customer reviews and generic labels that have no place in front of a client. Every visible word must be real or explicitly presented as provisional. A mock-up that fails on any of these three points is not ready: it needs human rework before it can be approved. This discipline is what separates a professional presentation from a mere technical demo.
The real question is no longer "AI or not"
Should you, out of caution, stick to a fully manual approach? The question hardly makes sense any more: with nearly three quarters of professional designers using AI every day, refusing it on principle means working more slowly than the market with no guarantee of working better. The point is not to use AI or reject it, but to frame it: to know what you delegate to it and what you keep under human control.
That is exactly what the gap between the 72% who have adopted it and the 32% who trust it without checking shows: the profession has taken up the tool without handing over its judgement. AI in Figma is a remarkable execution accelerator, as long as it remains what it is, a means serving an intention, never the intention itself.
Our brand identity and design service uses these tools to speed up your projects without sacrificing quality. Tell us about your project to discuss it.
Sources: Figma, 2025 AI report · Blog du Modérateur, 72% of designers have adopted generative AI