According to the Figma State of the Designer Report (2026), 91% of designers say AI tools improve the quality of their work - and 89% report working faster (Figma, 2026). Designers using AI also report 25% higher job satisfaction. The question is no longer AI or designer - it is: who uses AI to get better?
At Webnique, our first experiments started in 2023 - back then with a simple question: can AI do my job? After three years of hands-on experience, the answer is clear: AI can handle many sub-tasks in the design process. Our daily work regularly required finding or creating images, graphics, and illustrations - time-consuming tasks where we saw real potential for AI.
A quick state-of-play: Midjourney V7 has been the default model since April 2025 - with photorealistic output, 20-30% faster generation, and real video output (since June 2025). The V8.1 Alpha launched in April 2026 (Midjourney Updates, 2025). The tool we tested in 2023 and the tool that exists today share little more than a name.
What has fundamentally changed: prompt adherence. V7 interprets prompts far more precisely than earlier versions - fewer creative surprises, more control. What has not changed: Midjourney is still not the right tool for precise typography in designs, rigid layout grids, or pixel-perfect deterministic results. But for what it does well, it is better in 2026 than ever before.
Adobe Firefly has generated over 24 billion images since its launch - a clear signal that the market recognizes the value (Quantumrun / Adobe, 2026). For 3D concept visualizations and photorealistic images, AI delivered convincingly even back in 2023. Concepts that previously required hours of 3D modeling now emerge in minutes.
In our projects, we now use AI image generation primarily in the early concept phase: quick visualizations for client presentations, alternative visual directions for comparison, texture ideas for UI elements. What used to cost a full design day now happens in the kickoff session.
AI also surprised us with its ability to create visually compelling photos that often surpassed available stock solutions. Midjourney even had creative ideas we had not expected - for instance, it interpreted the word flowers in the context of a woman as her wearing a flower-adorned hat.
88% of creative professionals report working faster with AI - but what about the other 12%? Our experience was mixed when it came to AI-based illustrations meant to visualize complex content. Midjourney V7 is not the right tool for precise typography in designs, rigid layout grids, or pixel-perfect deterministic output. For those tasks, human design work remains irreplaceable.
The more concrete and technical the design requirement, the larger the gap between AI output and actual need. The more open and conceptual the task, the more useful AI becomes. Understanding this means deploying AI where it genuinely adds value.
McKinsey estimates that AI could boost design productivity by 40-70% by 2030 (McKinsey State of AI, 2025). In 2023, AI image generation was an experiment for early adopters. In 2026, it is a standard component of professional design workflows. Bitkom reports that 41% of German companies are already actively using AI - double the previous year (Bitkom, 2026).
The biggest takeaway from three years of AI image generation: the quality problem has been solved. The new challenge is prompt competency. The question is no longer Can AI do this? - it is: Can you tell AI what to do?
For commercial teams, Adobe Firefly is the safest choice - fully integrated into Creative Cloud, copyright-cleared, and proven with over 24 billion generated images. For artistic quality and stylistic range, Midjourney V7 remains the leading tool. Many professional teams use both in parallel.
AI can quickly generate first concepts and moodboards - but it cannot deliver pixel-perfect, implementation-ready UI layouts. Tools like Figma or Adobe XD remain essential for actual UI/UX design. AI-generated images work best as inspiration and early concept visualizations.
The progress is enormous: Midjourney V7 (April 2025) delivers photorealistic results experts can no longer distinguish from real photography. Added: video generation (June 2025), better prompt adherence, improved text rendering, and V8.1 Alpha (April 2026).
No - but they are fundamentally changing the work. According to Figma (2026), 91% of designers say AI tools improve their work quality, and AI users report 25% higher job satisfaction. AI handles repetitive image tasks and creates more room for strategy and UX.
Want to know how AI-powered image generation and UI/UX design can work together in your web project? We can help - with real experience from three years of AI in the design workflow.
