If you want the short answer: AI is currently an accelerator in website projects, not a replacement for the person designing the site. It changes the early stages the most (first drafts of copy, structural suggestions, routine code, and image preparation), while the decisions about what the site should achieve, how trust is built with your customers, and how it all fits your actual business are still made by a human. The practical consequence for small businesses: builds are faster and more predictable, but the gap between a mediocre site and a good one is wider than it used to be.
Below we go through what has genuinely been automated, where the limits are, how this affects cost and timeline, and what to watch for when commissioning a site in 2026.
What has actually been automated in website projects?
Three years ago, most of the work on a small website went into repetitive steps: writing first versions of the copy, assembling the same components over and over, adjusting images, checking everything on mobile. Today a good share of that takes a fraction of the time.
First drafts of content. AI tools quickly produce a skeleton for an about page, a service description, or a set of frequently asked questions. That is not finished copy, but it beats a blank page. The time that used to go into “how do I even start” moves to editing and fact-checking instead.
Routine code. Standard components such as forms, navigation, service cards and pricing tables get written far faster with assistance. The same goes for turning a design into code and for the small fixes that used to mean half an hour of digging through documentation.
Visuals. Illustrations, backgrounds, icons, and blog cover images no longer require an hour and a half in stock libraries. For photos of products, premises, and people, a real photographer is still the better call, something we cover in our step-by-step guide to building a website.
Checks and maintenance. Automated accessibility, speed, broken-link, and security-update checks now run quietly in the background. Problems are often caught before a visitor notices them.
Where does AI still not help?
The boundary is clearer than the headlines suggest. AI handles what has already been written or built somewhere. It handles poorly what only the business owner knows.
It doesn’t know why your customers actually call, which question they always ask before deciding, which three points win the job in your town, or what happened last time you raised your prices. It also doesn’t know what you can realistically deliver, and a site that promises something you can’t fulfil works against you.
That is why the first conversation with a client remains the most valuable part of a project. It surfaces the material AI has no way of obtaining: real numbers, examples, objections, the reasons people walk away. Without it you end up with a site that is neat, grammatically flawless, and entirely interchangeable.
The second thing AI doesn’t solve is judgement. One button here or three? Is a public price list an advantage or an obstacle? How much text is too much for someone reading on a phone during a break? Those are decisions with consequences, and they’re made by knowing the specific business.
What does a website project look like in practice now?
The order of the steps hasn’t changed, but the centre of gravity has. The first part is the conversation and gathering the facts: what you offer, to whom, at what price, what makes you different, and what customers ask most often. No tool speeds this up, and it now gets more attention than it used to.
Structure comes next: which pages exist, what goes on each one, and in what order a visitor moves through them. Here AI works well as a sounding board. It offers layout variants and flags the things that are easy to forget, like a terms page or a plain statement of your service area.
Then the build. Copy is written in two passes: a first draft appears quickly, then gets rewritten with real client-supplied detail. Code is written with assistance, but every component is manually checked on mobile, across browsers, and with a keyboard instead of a mouse.
Finally the pre-launch checks: speed, readability, working forms, metadata, and analytics. That part is more thorough than it used to be, simply because there is more time left for it.
Does that mean a website is cheaper now?
Partly yes, but not as much as people expect. What got cheaper is the part that was always mechanical. What got more valuable is the part AI doesn’t cover.
In practice: a simple one-page site is built faster than it was two years ago, and that shows in the price. More complex projects with multiple services, two languages, or an online store haven’t become dramatically cheaper, because the bulk of the work there isn’t typing code but decisions, content, and integrations. Current ranges and timelines are published openly on our pricing page.
What has changed is delivery speed. Timelines that used to run three to four weeks are genuinely shorter now, and mid-project changes cost less because they’re quicker to implement.
How do you spot a site AI built on its own?
There are a few recognisable signs, useful both when assessing someone else’s quote and when reviewing your own site.
The copy reads smoothly but says nothing measurable. No numbers, no timelines, no place names, no examples. The listed advantages could be claimed by any company in the same industry. There is an FAQ section, but it answers questions nobody asks. The images are attractive and have nothing to do with what you do.
By contrast, a site that works usually contains specifics: exactly what you deliver, in what timeframe, at what price or range, who has already worked with you, and what happens after someone sends an enquiry. AI can help shape that content, but it cannot invent it.
What does this mean for you as the client?
Three practical takeaways.
Ask what in the quote is human work and what is tooling. Using AI isn’t the problem. Being charged a week’s rate for an AI output is. A transparent quote says clearly where the hours go.
Invest in content, not only design. Design is comparatively easy to obtain now. Content that answers your customers’ real questions remains the most expensive and most useful part of the project.
Insist that someone verifies the facts. Prices, timelines, legal obligations, and technical details must pass a human check before publication. An error in a price list costs more than the time it saved.
If you’re considering a new site or a redesign, have a look at how we approach web development, and if you want a concrete estimate, fill in the short questionnaire and we’ll come back with a proposal that states what we do and what the tooling does.
Conclusion
AI hasn’t eliminated website work, it has shifted it. Less time goes into production, more into decisions. For small businesses that’s good news: a good website is more accessible than ever, provided someone still asks the right questions before the building starts.