AI-assisted design is changing how creative teams move from idea to execution.

Every design team I talk to right now is somewhere in the middle of working out what AI actually means for how they work. Plenty of tools. Plenty of impressive demos. Plenty of opinions about where it's all heading.

The instinctive move is to bolt AI onto the existing process. New tool, tell the team to experiment, see what sticks. Nine times out of ten that just produces more output. Not better work. Just more of it.

I think if you want real value from these tools, parts of the workflow need rethinking.

Where AI genuinely earns its place

The biggest win is speed. Specifically, speed from idea to something you can actually put in front of people and argue about. Early concepts, moodboards, rough layouts, copy options, image exploration, the repetitive corners of production. An hour now, where it once took a day or two.

That's not just a time saving. It changes the conversation.

Instead of describing an idea in the abstract, you can put three or four real directions in front of a client and have a far more useful conversation about what's working.

That's where AI earns its keep. Not producing the finished piece. Getting everyone to the interesting part of the process, sooner.

Where you need to be careful

The problem shows up when AI-generated work starts looking finished before it is.

These tools are unnervingly good at producing something convincing. Clean layout, plausible copy, just enough polish that a room full of sensible people nod and think you're nearly there.

Often you're nowhere near there.

The missing bit is usually what makes a piece of work feel like a specific brand rather than a competent generic one. An accessibility problem. A user journey that doesn't survive contact with a real customer. Or simply that nobody's actually thought about it. The work just looks thought about.

Easy to miss, when what's on screen already looks 90% done.

Which is why review can't be a step bolted on at the end. If AI's in the workflow, judgement has to sit in the workflow too, not get tacked on afterwards.

What a good workflow actually looks like

Use AI to widen the exploration and speed through the repetitive stages. Keep hard checkpoints where someone with real experience stops and asks whether the work is actually right. Right for the brand. Right for the audience. Right for the business. Right for the specific problem in front of you, not a generic version of it.

There's a real difference between a team using AI to produce more work and a team that's changed how it works to get more from AI. The first leaves you with a bigger pile to review. The second leaves you more time to think.

That's the actual opportunity. I'm not interested in AI making designers type faster. I'm interested in it clearing away the less valuable work so there's more room for the things that still need experience, judgement and a bit of imagination.

I've always embraced new technology and used it to make the work better. But the craft still comes first. It always will.

Discover the prev article in this series