Will AI replace web designers?

The short version

The parts of web design that were always mechanical — laying out a standard page, producing five variations of a hero section, writing serviceable placeholder copy, converting a design into working code — are getting cheap very quickly. That is genuinely happening, and pretending otherwise helps nobody.

The parts that were never mechanical are not moving: deciding what a business should say, knowing which of five options is right for this client, and taking responsibility when it is wrong.

What AI does well now

First drafts of layout. Structure, spacing, a competent arrangement of standard sections.

Volume. Ten variations in the time it took to make one. This changes how you work more than what you charge for.

Boilerplate code. The unglamorous scaffolding that took hours and impressed nobody.

Serviceable copy. Not good copy — serviceable, which is enough to review and correct rather than write from nothing.

Together, these compress the middle of the job substantially. A build that took three weeks can take one.

What it cannot do

Know the business. Not the industry — this business. That the owner wants to grow the catering side, that half the customers arrive from a station, that the thing everyone asks about is not on the menu. None of this is inferable, and all of it decides the design.

Choose. A model will produce five directions. Knowing which one fits this business, this town, this budget is judgement, and judgement comes from having been wrong before.

Say no. The most valuable thing a good designer does is talk a client out of the thing they asked for. No generator does this — they optimise for compliance.

Be accountable. When the booking form fails on a Saturday, somebody has to answer for it. That responsibility is what people are paying for, and it cannot be automated.

What actually changes

Not "designers are replaced". The shift is in what the fee is attached to.

Away from production. Charging by the page or by the hour of assembly gets squeezed, because assembly is what got cheap.

Toward judgement and outcomes. What should this site do, what should it say, and does it work. Those hold their value.

Toward the relationship. Ongoing work — maintenance, content, advice, being the person who answers — becomes a larger share of a sustainable practice, because it is the part a generator cannot occupy.

The designers who struggle will be the ones whose entire offer was production. The ones who do well will be the ones who were always selling judgement and using production to deliver it.

The honest advice

Use the tools. Refusing on principle just means doing slowly what a competitor does quickly, and clients do not pay extra for that.

Then spend the time you saved on the parts that are still hard: understanding the business, choosing well, saying no when it matters, and being reachable when something breaks. That is the work that was always worth paying for — the rest was overhead we were charging for because it took time.

Where this leaves people who build sites for a living

Worth being concrete rather than reassuring, because "the work moves up the stack" is easy to say and hard to bank.

The commercial shape that survives is the one where you are paid for judgement and for the relationship, rather than for hours of production. In practice that means fewer one-off builds and more ongoing arrangements: you choose well, you set it up properly, and you stay reachable.

That is the shape Helm's partner programme is built around. You build the site — or have the Helm team or Helm AI build it — and the client gets a branded dashboard for the day-to-day, so they are not queuing behind you for a price change and you are not billing an hour to make one. You keep the relationship and a recurring share of it instead of chasing the next build. The arithmetic is in recurring revenue for a freelance web developer.