AI can build a website.
That sentence would have sounded like science fiction not very long ago. Today it is an advertisement.
Give an AI tool a description of a business and it can produce layouts, write headings, suggest colours, generate images and create working code in minutes. With the right tools and enough prompting, it can assemble something that looks remarkably like a finished website.
So there is an obvious question for somebody considering a new site in 2026.
Why hire a web designer?
It is a fair question. It is also much more interesting than pretending AI is a gimmick that serious designers can safely ignore.
We use AI ourselves. A lot.
Just not for the reason people sometimes assume.
AI is extraordinarily useful
Let's get this part out of the way first.
AI is not coming for web design at some mysterious point in the future. It is already part of modern web development.
We use it for difficult code analysis, tracing bugs through large projects, working through data, researching technical problems, exploring implementation approaches and accelerating repetitive development work.
On a complicated migration, for example, AI can help reason through thousands of references and relationships far faster than a human could inspect them manually. When a plugin has grown into a substantial codebase, it can help trace how one change travels through multiple files and systems. When an obscure error appears, it can help narrow the search.
That is enormously valuable.
There are days when it feels less like having a tool and more like having an extremely fast technical colleague who has read every manual ever written but occasionally needs to be stopped before confidently walking into a cupboard.
The important part is knowing when to stop it.
Generating something is not the same as understanding it
AI is very good at producing plausible answers.
Web design is full of situations where plausible is not enough.
A business owner might ask for a bigger banner. The real problem may be that the page has no hierarchy. They might ask for more information above the fold when the real problem is that nobody knows which information matters. They might want their competitor's layout when their own customers, services and positioning are completely different.
A designer's job is not simply to execute the sentence that was typed into the prompt box.
It is to understand what somebody is actually trying to achieve.
AI can generate a website. A designer has to understand why it should exist.
That distinction becomes more important as generation gets easier.
The hard part was never drawing the rectangle
Website builders have been making web design easier for years.
Before AI, there were templates. Before increasingly sophisticated visual builders, there were page builders. Before those, there were themes. Every generation of tools has reduced the amount of technical work required to put text, images and buttons onto a screen.
And yet businesses still hire designers.
That is because the valuable part of design is not knowing how to create a rectangle with rounded corners.
It is deciding whether the rectangle should be there at all.
What should somebody see first? What does the business need to communicate before asking for an enquiry? Which project proves the claim being made? Should this page feel authoritative, technical, luxurious, approachable or urgent? Is the navigation organised around the company's internal departments or around the way customers actually think?
Those decisions are connected to a business, not just to a browser.
AI has taste. Unfortunately, it has everybody's taste.
Ask a capable AI system for a premium modern website and there is a good chance it can produce one.
There may be an enormous headline. There may be a dark background. There may be a tasteful gradient, a glassy card, some generous spacing and a button with a reassuringly contemporary border radius.
It may look very good.
It may also look strangely familiar.
AI learns from enormous amounts of existing material. That makes it excellent at recognising patterns and combining them. It also means that without strong direction, it naturally gravitates toward solutions that already look like successful solutions.
A designer can use those patterns too. The difference is being able to decide when the familiar answer is right and when the business needs something else.
Taste is partly knowing what works.
It is also knowing what to leave out.
Context is where the real project lives
A website rarely exists by itself.
It may need to preserve years of search visibility. It may connect to an existing CRM. It may inherit thousands of old URLs. It may need to work with a client's hosting environment, email system, product database, analytics, forms or internal workflow.
It may also need to account for something no prompt could possibly know until somebody asks the right question.
This is where real projects become much less like generating a homepage and much more like solving a system.
We saw this very clearly during our migration of Steven Murphy Electrical Contractors from Kentico to WordPress. The visible redesign was only one part of the job. Underneath it sat hundreds of pages, media, attachments, aliases, internal references, existing search history and enquiry processes that all had to survive the move.
An AI tool could help us analyse and solve pieces of that problem. It could not be responsible for deciding what was safe to change on a live system, what had to be preserved, what needed human verification or when we should refuse to take a shortcut.
Responsibility is not a feature you can prompt into existence.
There is also the small matter of being wrong
AI can be confidently wrong.
So can humans, obviously. We have had thousands of years of practice.
The difference is that AI can produce a wrong answer with extraordinary speed, excellent grammar and enough technical detail to make the mistake look impressively qualified.
That matters in development.
A suggested code change can fix the symptom while introducing a subtler problem elsewhere. A generated SEO recommendation can sound sensible while being based on an outdated assumption. A migration script can be syntactically perfect and conceptually disastrous.
This is why we treat AI output as something to inspect, test and verify, not as an instruction from the mountain.
The better the tools become, the more valuable judgement becomes too.
AI makes experienced people faster
This is the part of the AI conversation we find most exciting.
The strongest combination is not human versus AI.
It is an experienced human using AI well.
A developer who understands WordPress can use AI to move through a WordPress problem faster because they recognise when the proposed solution makes sense. A designer can explore more possibilities because they know which ones deserve to survive. Someone experienced with SEO can use AI for research and analysis without assuming every confident paragraph is current advice.
The tool increases leverage.
But leverage works in both directions.
If you know what you are doing, AI can help you do considerably more of it. If you do not know what you are doing, AI can help you do considerably more of that too.
So what are you actually paying a designer for?
You are not paying someone because they possess secret knowledge of how to put a picture next to some text.
You are paying for judgement.
You are paying for someone to understand the business before deciding how it should be represented. To recognise when a request will damage the result. To notice that the real problem is somewhere other than the place everybody is currently looking.
You are paying for design decisions, technical decisions and the experience to know that those are often the same decision.
You are paying for somebody to take responsibility for the result.
And, if the relationship is a good one, you are paying for somebody who will still understand the website six months later when something changes and you need help.
That last part is remarkably difficult to generate with a button.
Will AI replace some web design work?
Almost certainly.
There are websites that never needed a custom designer in the first place. A simple temporary site, personal project or basic information page may be perfectly well served by an automated builder.
That is not a failure of the design industry. It is a better tool being matched to an appropriate job.
AI will also automate parts of professional web design that previously consumed human time. We welcome that. Nobody entered this profession because they dreamed of spending Wednesday afternoon manually comparing 400 URL references.
The interesting result is that the human part of the job becomes more concentrated.
Understanding. Direction. Taste. Strategy. Verification. Communication. Responsibility.
The less time we spend doing work a machine can perform reliably, the more time there is for the parts that actually need us.
The future probably isn't human versus AI
We do not think the winning approach is to reject AI and insist that every line of code must be lovingly typed by candlelight.
Nor do we think the future is a prompt box producing an endless supply of interchangeable websites while everybody congratulates themselves on efficiency.
The useful middle is much more interesting.
Use AI to analyse the difficult code. Use it to accelerate research. Use it to find patterns in data. Use it to help debug the maddening problem that has survived three cups of coffee. Use it to remove repetitive work and expand what one experienced person can accomplish.
Then use human judgement to decide what is true, what is useful, what is appropriate and what is actually worth building.
Human plus AI plus experience is a much more powerful proposition than either human or AI used badly.
So yes, AI can build a website.
The question is no longer whether it can.
The question is whether the website understands the business, earns trust, solves the right problems and continues to work when the easy part is over.
That is still design.
