We need to talk about the AI-accelerated workflow. The standard advice has turned into “let the machine do it,” and the quality of what we ship is dropping because of it. I have been in the WordPress and UX ecosystem for over 14 years. Tools come and go, but handing intent over to a generative model without checking the output builds technical and ethical debt fast.
When generative AI first landed in my local environment, I was uneasy and curious at once. An interface appearing in seconds, with sensible spacing and typography, raises a fair question: If a machine can do this, where does that leave me? Then came a few dozen hallucinated layouts and inaccessible color palettes. AI is not replacing the designer. It raised the floor on production.
What AI does better: the boring production work
AI is better than we are at parts of design work, mainly speed and volume. Instead of spending three hours sketching three concepts, you can review thirty in three minutes. A report by McKinsey puts the time saved on creative tasks at up to 70%.
- Consistency: Design systems only work when the rules get followed every time. AI does not eyeball it, and it does not get tired of color tokens and spacing scales.
- Data processing: Reading thousands of heatmaps or scroll-depth sessions is fast work for a machine. It tells you what happened.
- Ideation: Cheap variations, as many as you want. I went through how that plays out in Pragmatic AI Workflow Automation.
Where human strategy still does the work
AI has one bottleneck it will never clear: it is not human. It can imitate empathetic language, but it does not feel the quiet rage of a broken checkout form or the worry of handing over sensitive data. That gap is why human strategy decides whether an AI-accelerated workflow produces anything worth shipping.
UX work is mostly navigating ambiguity and arguing for people inside systems built for efficiency. AI optimizes whatever objective you hand it. Ask for engagement and it will happily build you a slot machine.
Ethical design needs somebody willing to say “We could do this, but we shouldn’t.” A model does not recognize a dark pattern or a variable reward loop for what it is. Strategy also lives in context, and AI does not sit in stakeholder meetings or track the organizational politics that decide how long a product actually survives.
From maker to director of intent
Designers are moving from making outputs to directing intent. A film director does not operate every camera or build every set, but the story and the emotional impact are still theirs to answer for. In an AI-accelerated workflow, your judgment is the thing being paid for, not your ability to draw buttons.
I have watched senior designers spend less time inside Figma and more time facilitating workshops and protecting user needs. When options are cheap and plentiful, discernment becomes the scarce skill in the room. Who is this for? What happens when it fails? Who does it disadvantage? Those questions do not answer themselves.
If this AI-accelerated workflow business is eating your dev hours, hand it over to me. I have been wrestling with WordPress since the 4.x days.
The designer’s accountability
AI will not take your job, but a designer who knows how to direct it might. That makes us more accountable for what ships. Bad design used to come with excuses attached: not enough time, not enough data. Those excuses are going away. Once execution stops being the hard part, the strategy is what you answer for.
Start small. Try something like secure AI connectors in your own workflow, and keep putting hours into psychology, communication and ethics. The human part of the job is not shrinking; it just asks for more deliberate choices. Ship it.
