AI is a tool. Not a replacement. Not a shortcut.

Every creative industry is having the same conversation right now. Should we use AI? Will it replace designers? Is this cheating? These are the wrong questions. The better question is: where does AI make the work better, and where does it make it worse?

The studios getting the most out of AI are not using it to skip the thinking. They are using it to accelerate the parts of the process that eat time without adding creative value, while keeping the strategic and craft decisions firmly in human hands.

Here is what we have learned from working with it.

Where AI earns its place

Concepting and ideation. AI image generators like Midjourney and Adobe Firefly are genuinely useful for rapid visual concepting. You can explore 40 rough directions in an hour rather than a day. That speed changes how you approach early-stage work. You can be more ambitious with your initial exploration and narrow down faster.

Mood boards and visual references. AI is excellent at producing visual references for internal use. Showing a client a direction before committing to design time is valuable. Just be clear that what you are showing is a reference, not a deliverable.

First-draft copy. AI can write a serviceable first draft of almost anything. It will not be good copy, but it will be a starting point faster than a blank document. For social captions, headlines, or rough long-form structure, it saves time. A human still has to shape it into something with real voice and intent.

Content variations at scale. If you need 20 variations of a headline or caption for an A/B test, AI gets you there in minutes. That is time that used to disappear into manual iteration.

Research and brief preparation. AI tools are strong at summarising research, gathering competitive context, and helping structure a project brief. Not infallible, but useful as a starting point.

Where AI will let you down

Print production. This is the big one. AI-generated images are screen outputs. They are RGB, typically 72dpi, with no bleed, no crop marks, and no colour profile suitable for print. Print requires 300dpi minimum and CMYK colour values. Converting an AI image to CMYK for print will shift the colours in ways you cannot predict or control. Blues go muddy. Greens flatten. What looked clean on screen looks wrong on press.

There is also no way to control bleed areas or file structure through an AI image generator. Print production requires precision that these tools are not built to deliver. Do not try to use AI image output as production-ready print artwork. It is not.

Typography inside images. Every major AI image tool still handles text poorly. Letters are wrong, spacing is uneven, words are misspelled or distorted. Even after significant improvements in recent updates, you cannot rely on AI to render readable type inside an image. Always add type in your design software after generation, never inside the prompt.

Brand consistency. AI cannot maintain your brand colours, typefaces, or logo accuracy across outputs. It does not know your Pantone references or your hex codes. Every generation is a new interpretation. This makes AI unsuitable for producing brand-compliant assets without significant human correction after the fact.

Commercial image rights. Many AI image tools have trained on images scraped from the web without explicit licensing. Getty Images pursued legal action against Stability AI in 2023 over exactly this issue. For commercial client work, Adobe Firefly is the most defensible option because it was trained exclusively on licensed Adobe Stock content. Know what your tool was trained on before you deliver AI-generated work to a client.

Prompting is a skill, not a feature

The gap between a mediocre AI output and a useful one is almost always in the prompt. Vague inputs produce generic outputs. The more specific and informed your prompt, the closer the output gets to something you can actually use.

Good prompting for creative work means referencing specific aesthetics, photographers, design movements, lighting conditions, colour palettes, and compositional structures. "A modern brand identity" produces noise. "A minimal Swiss-style brand identity, black and white, geometric sans-serif typography, high contrast, editorial" produces something you can work with.

Negative prompts matter too. Telling the AI what to exclude, "no people, no gradients, no photorealism," shapes the output as much as what you ask for.

According to a 2023 McKinsey report on generative AI, the most effective users in creative functions spent significant time learning to prompt well before seeing meaningful productivity gains. The tool does not work without the skill to direct it. That skill takes time to develop and is worth developing properly.

Think of prompting the same way you think of briefing. The quality of what comes back is directly proportional to the quality of what you put in.

You have to understand the product before you can brief the AI

This is the part that gets missed most often. AI is very good at executing what you ask for. It is not good at filling in gaps in your own understanding of what you need.

If you do not understand who the brand is for, what feeling the design needs to create, what the product actually does, or what market it is competing in, no AI tool will figure that out for you. It will generate confident, polished output that answers the wrong question.

The studios that get the best work out of AI are the ones who do the thinking first. Strategy, audience clarity, creative direction, all of that comes before the AI enters the process. Then AI accelerates the execution of a direction that already makes sense.

This is not a limitation of the technology. It is how good creative work has always worked. A junior designer who does not understand the brief will produce the wrong work no matter how skilled they are. AI is no different. The brief still has to be right. The strategic thinking still has to happen. AI just moves faster once it has a clear direction to follow.

Use it like a skilled assistant, not a creative director

The studios that are losing ground to AI anxiety are the ones treating it as something that replaces judgment. It does not. It replaces time-intensive tasks that did not require much judgment in the first place.

At klein studio, we use AI for ideation, research, and first-pass content. We do not use it for print production, brand-critical assets, or anything that requires precise typographic control. We prompt carefully, evaluate critically, and always put human craft on top of anything AI generates before it gets close to a client.

The creative value we offer is not in our ability to produce images fast. It is in knowing what image to produce, why, and for whom. AI makes the production faster. It does not make the thinking easier. And the thinking is still the work.