Artificial intelligence is fundamentally changing the work in web and graphic design. A lot has shifted for me, too, over the past few months. Some tasks I now handle completely differently than just a few years ago. Others deliberately remain handcraft. And then there are projects where AI takes a lot of work off my hands but is far from deciding everything.
Graphic design: fewer classic jobs, more room for special ideas
Classic flyer and poster jobs have become much rarer for me. Of course, there are still clients for whom I design print products. These include, for example, newspaper ads and banners for larger companies that rely on classic outdoor and print advertising and want to keep using it in a targeted way.
With projects like these in particular, a reliable final check is crucial. Colors have to be right, formats have to fit and ads should adapt as flexibly as possible to different requirements. An ad might have to run in a different format at short notice, or a design element needs to move to a different position.
AI can help here, too. For one of my clients, for example, I have already set up a template with the corporate design in Claude Code. The system keeps giving me interesting approaches from which new design ideas emerge. The actual final artwork and production still happen in Adobe InDesign, though. Why? Because I want to stay in control. I want to be able to move a piece of type quickly, adjust spacing precisely or prepare a layout for another format without detours.
I don’t want to prompt every design decision. I want to keep mastering my craft. Because if we rely too much on an AI doing everything for us, there’s a risk that we lose certain skills over time. For me, it’s therefore part of the job to keep designing myself, experimenting and actively using technical skills. AI should expand my work, not replace my design thinking.
It’s different with individual projects. When it comes to art, unusual client requests or ambitious museum and exhibition posters, design gets particularly exciting. Here it’s not just about arranging information as efficiently as possible. It’s about attitude, atmosphere, emotions and a visual language of its own.
That’s exactly what takes new ideas and creative impulses. AI can combine existing approaches, recognize patterns and make surprising suggestions. Much of it, however, is a remix of influences that already exist. Human creativity partly works in a similar way, but it also brings in personal experience, emotions and individual perspectives. And it’s exactly this mix that makes good design for me.
Web design: I no longer design every subpage in advance
My workflow has changed much more in web design. I’ve now reached the point where I have websites developed directly in HTML with Claude Code. I set the design direction and have a working website emerge from it.
The starting material can look very different. Sometimes there’s already a finished design. Sometimes it’s just individual design elements, a few buttons, certain typography rules or a clear idea of how the website should feel. On that basis, we develop the design and the technical implementation step by step. I give feedback, have elements changed, test different approaches and work my way toward the desired result together with the AI.
What I no longer do in every project: fully work out all subpages in a classic design process before development even begins. In many cases, that no longer makes sense to me.
Typical elements of websites and apps already follow established UX principles. How big should a button be? How do you design a form that’s easy to understand? How do font sizes and spacing need to work together? How does the layout change on mobile devices? Which breakpoints make sense? These are questions for which AI can already deliver good starting solutions. Of course, I have to check, adapt and optimize these results. But I no longer have to design every single component manually from scratch before I can start on the actual website.
What used to take many hours for wireframes, layouts, responsive variants and working out all the subpages can now flow directly into development. The result isn’t just a finished design. It’s a website that has already been built.
Fewer handovers, but not necessarily less work
A common misconception is that AI projects automatically mean less working time. For me, that’s not necessarily the case. I still invest about as much time in a good web project as I used to put into concept and design. The difference lies in what exists at the end of that work.
In the past, an extensive design process first produced a design concept that then had to be handed over to development. Today, the working website emerges in parallel with the creative process. In many projects, this eliminates a considerable part of the classic development work and the handovers that came with it. Depending on scope and starting point, this can save around 30 to 40 percent of previous project costs, for example. That’s not a flat figure, though – it depends heavily on the individual project.
Because AI-assisted development costs money, too. I deliberately work with powerful models, for example Claude Opus, and invest in additional credits for demanding projects. When I work on a website for several days, relevant usage costs add up quickly. For a large project, that can easily be 300 euros just for additional AI credits.
You can’t see these costs in the finished website. The client sees the result: a website that is individually designed, stands out from many other sites and has already been built. The credits are still part of the project costs and have to be factored in accordingly. So what matters to me is not spending as little money on AI as possible. What matters is using the available tools sensibly and creating a result that genuinely justifies the effort.
A good website doesn’t come from a single prompt
Anyone who thinks you only need to briefly describe a website to Claude Code or another AI tool to get a truly well-thought-out result will quickly hit limits on more demanding projects. A convincing result comes from many small decisions and an ongoing exchange.
What content does the website need? How are the services presented? Which subpages are necessary? Which information should be findable by search engines? How do texts need to be written to suit the target audience? And how do design, user guidance and technical implementation interlock?
Writing and revising the texts is part of it, too. I work with both Claude and ChatGPT. The content basis comes from the client: their services, their goals, their target audience and their knowledge of their own business. From that, we develop texts together, check them, have them proofread and revise them until they fit. The client remains an important part of the process. The AI makes suggestions and helps with implementation. But the responsibility for the content and the decisions doesn’t disappear.
For a really good website, this process easily takes three weeks. Sometimes longer, depending on scope and complexity. Because a website isn’t just a collection of nice pages. It has to work as a whole.
What remains is the human eye
With all the technical development, there’s one point that matters most to me: I don’t want to just operate AI. I want to understand what I’m designing. I want to be able to follow decisions and still be able to do things myself. In graphic design in particular, I will therefore keep working with classic tools like InDesign whenever they’re the better solution for the task.
And in web design, using AI doesn’t mean that I lean back and wait for the result. On the contrary: I accompany the entire process, question suggestions, optimize details and make sure that a technical implementation becomes an individual website.
There’s another part of my work on top of that: photography. As a photographer in Dresden, I can offer my clients authentic pictures made specifically for their business. These pictures flow directly into the design and make sure that a website doesn’t just work technically, but also comes across as personal and credible. Because a good website doesn’t consist only of clean code, suitable font sizes and a well-thought-out mobile layout. It also lives from the people behind a business, from real insights and from imagery that suits the brand.
My conclusion: AI is changing the craft, but it doesn’t make it obsolete
Work in web and graphic design is changing enormously right now. Some tasks that used to take me many hours can now be solved differently and much more directly. Others deliberately remain in my hands. For me, the future doesn’t lie in automating as much as possible or replacing every single step with a prompt. It lies in choosing the right tools for each task.
AI can speed up development processes, deliver new ideas and open up possibilities that would have been much more laborious just a few years ago. But it needs clear guidelines, creative impulses, professional oversight and people who know what they want to achieve. That’s exactly why I don’t see this development as the end of the classic design profession. I see it as a change in the craft.
Less time for repetitive work. More room for individual solutions. And still a human who decides what is really good in the end. That, for me, is modern web and graphic design from Dresden.
A website or design for your business?
How I build websites and what they cost is on the web design page; for logos, posters and corporate design, there’s the graphic design page. How else I use AI in everyday work is described in the posts AI and my profession and Where do I actually stand on AI-generated images?