What AI Actually Changes About Web Design (and What It Doesn’t)

Every few months another client asks me whether AI is going to make web design obsolete, or whether they should just let a tool build their whole site. The honest answer is that AI has changed parts of my job a lot, and other parts not at all. Here’s what I actually see when I build sites for small businesses around Commack and the rest of Long Island.

What AI is genuinely good at now

I’m not going to pretend it’s useless. In 2026, AI handles a real chunk of the grunt work that used to eat hours.

• First-draft copy. Give a model your services, your town, and a couple of real customer questions and it’ll produce a usable first draft. It won’t be good enough to publish, but it saves the blank-page problem.
• Layout scaffolding. Tools like v0, Claude, and Cursor can turn a description into working HTML and CSS in seconds. I still rewrite most of it, but starting from something beats starting from nothing.
• Image cleanup and generation. Background removal, upscaling, and generating placeholder hero images are basically solved.
• Code explanation. When I inherit a WordPress theme or a mess of legacy jQuery, pasting it into a model to get a plain-English summary is faster than reading it line by line.
• Accessibility and SEO audits. Models are decent at spotting missing alt text, poor heading order, and contrast problems if you ask specifically.

What it still can’t do

This is the part the “AI will replace designers” crowd skips.

AI doesn’t know your business. It doesn’t know that your highest-margin service is the one you hate doing, or that your customers call instead of filling out forms, or that the last redesign tanked leads because the phone number moved below the fold. Those are judgment calls that come from talking to you and looking at your actual traffic.

It also can’t be trusted with facts about you. I’ve watched models invent awards, fabricate review counts, and confidently write “family-owned since 1987” for businesses founded in 2019. If you publish AI copy without checking every claim, you’re publishing lies with your name on them.

And it can’t taste. Generated layouts tend to converge on the same look: big rounded cards, purple gradients, centered hero text. It reads as generic because it is generic. Standing out still requires a human deciding what to leave out.

A real example: the schema markup problem

Local SEO for a Commack or Huntington business leans heavily on structured data. Ask a model to write LocalBusiness schema and you’ll usually get something like this:

<script type="application/ld+json">
{
 "@context": "https://schema.org",
 "@type": "LocalBusiness",
 "name": "Your Business",
 "address": "123 Main St"
}
</script>

It’s not wrong, it’s just incomplete in ways that matter. No telephone, no geo coordinates, no openingHoursSpecification, no areaServed. Google can’t use what isn’t there. Here’s a version that actually does something:

<script type="application/ld+json">
{
 "@context": "https://schema.org",
 "@type": "LocalBusiness",
 "name": "Example Plumbing",
 "telephone": "+1-631-555-0100",
 "address": {
 "@type": "PostalAddress",
 "streetAddress": "123 Main St",
 "addressLocality": "Commack",
 "addressRegion": "NY",
 "postalCode": "11725",
 "addressCountry": "US"
 },
 "geo": {
 "@type": "GeoCoordinates",
 "latitude": 40.8429,
 "longitude": -73.2926
 },
 "areaServed": ["Commack", "Smithtown", "Huntington", "Hauppauge"],
 "openingHoursSpecification": [{
 "@type": "OpeningHoursSpecification",
 "dayOfWeek": ["Monday","Tuesday","Wednesday","Thursday","Friday"],
 "opens": "07:00",
 "closes": "18:00"
 }],
 "url": "https://example.com"
}
</script>

A model can generate that, but only if you tell it every field, and only if you verify the coordinates and hours are real. The AI is the typist. You’re still the one who knows the answers.

Where AI quietly hurts small business sites

The bigger risk isn’t bad code. It’s sameness and bloat.

• Generic copy that ranks for nothing. If your service page reads like every other AI-written service page in your industry, Google has no reason to prefer it. Specific detail (real project photos, actual neighborhoods, named materials) is what separates you.
• Page weight. AI code generators love to pull in a framework and a component library for a five-page site. I’ve seen 3MB pages built for a plumber. That’s a slow site on a phone, which is where most of your traffic is.
• Fake confidence. Models will write about services you don’t offer because they pattern-match on your industry. Read everything before it ships.
• Duplicate content across your own pages. Ask for five location pages and you’ll get five near-identical pages with the town name swapped. That’s a thin-content problem, not a local SEO win.

What I actually do differently now

My workflow in 2026 looks roughly like this:

• Talk to the client first. Always. No tool replaces a 30-minute conversation about who their customers are and what a “good” lead looks like.
• Use AI for scaffolding, boilerplate, and first drafts, then throw away most of it.
• Write the copy myself or heavily edit it, because the details that make it convert come from the client, not the model.
• Keep the stack boring. Static HTML, a little CSS, maybe a light framework if the site genuinely needs one. Faster sites still win.
• Test on a real phone on a real cell connection, not just in a browser resize.

If you want a sense of how I approach this for local businesses, my Long Island web design work is mostly small business sites where speed, clarity, and getting the phone to ring matter more than looking trendy.

Should you just use an AI site builder?

For a personal project or a one-page site you’ll update twice a year, sure. Wix, Squarespace, and the newer AI builders are fine and cheap. For a business where the site is supposed to bring in calls, the tool isn’t the bottleneck. Knowing what to put on the page is. If you don’t have that clarity, an AI builder will just help you publish the wrong thing faster.

FAQ

Will AI replace web designers?

It’s replacing the parts of the job that were already the least valuable: repetitive coding, boilerplate, basic copywriting. It’s not replacing the judgment about what a specific business needs, and that judgment is most of what clients actually pay for.

Can I use AI-written content on my site without hurting SEO?

Yes, if it’s accurate, specific, and edited by a human who knows the business. The problem isn’t that content was AI-assisted. It’s that it’s generic, unverified, and duplicated across dozens of near-identical pages.

Is AI-generated code safe for a business website?

Often it works, but it frequently ships with unnecessary dependencies, outdated patterns, and security gaps nobody checked. Have someone who writes code review it before it goes live, especially anything handling forms, payments, or customer data.