🎉
Revolution wins Sustainable Business of the Year 2026!
Revolution wins Sustainable Business of the Year 2026!
Something has shifted in the opening conversation of most website projects. Clients now arrive familiar with AI design tools: they have watched a landing page appear on Lovable in ten minutes, seen a Figma page assemble itself on Claude Design, and read the trend pieces about how AI is going to compress software development timelines by an order of magnitude.
They arrive, reasonably, expecting that a website should now take a fraction of the time it used to and cost a fraction of the money.
The 2026 Web Designer Survey by 20i, based on responses from 500 web designers, captures the shift with two numbers. 90% of designers say clients underestimate the true cost of web design. Two-thirds report that managing client expectations has become significantly harder in the past year. The survey attributes much of this to the rise of AI-generated promises of “instant websites” and automated content solutions that have reshaped what clients think a project should involve.
The expectation misses one thing. AI has genuinely accelerated web work, but it has accelerated the parts that were never the slow parts.
The elements of a web project that AI has made dramatically faster are, mostly, the ones that used to take a day and now take an hour. This is a real improvement. It is also a smaller share of a full project than the demos suggest.
Initial design exploration. Showing a client three or four visual directions used to require a designer spending several days. Now it can happen in a single afternoon.
Wireframes and layout drafts. Getting a coherent structure for a homepage or landing page is meaningfully faster.
Copy first drafts. Headlines, subheads, and body copy that need refinement but give the client something concrete to react to.
Individual sections. A hero, a features grid, a testimonials block: all can be drafted in a few prompts.
Small code and utility work. Not the main build, but the bits and pieces around it.
Adoption of these tools is now the norm rather than the exception. Industry research reports that around 93% of web designers use AI tools in their workflow, and roughly 84% of developers say AI has improved their productivity. The gains are real. They just tend to concentrate in a specific set of activities.
Design work as a whole typically accounts for around 30-40% of a project’s total timeline. Even generous assumptions about AI compressing the design portion by half mean the overall project shrinks by roughly 15-20%. Useful, but nothing like the “one afternoon” narrative.
The genuinely time-consuming parts of a web project have not been meaningfully compressed by AI.
Understanding the business. A useful website is one that supports a specific business, aimed at a specific audience, pursuing specific goals. Working out what those are, what the visitor journey should look like, and what “success” means for the site cannot be prompt-generated. It happens through conversations, discovery sessions, and questions the client has often not been asked before.
Securing the content. Real photos of real people. Product descriptions written from actual knowledge. Testimonials from real customers. Case studies with real numbers. Getting this material out of a client and into a usable form is consistently described as the most underestimated phase of any web project. Having content ready before development begins can accelerate a project by 25-40%. Not having it ready delays the same project by the same amount. AI can polish drafts and generate placeholders, but the raw material has to come from the client, and getting it usually takes weeks rather than minutes.
Structural validation. Testing whether the proposed page structure actually works: whether every call to action leads somewhere meaningful, whether the URL structure makes sense, whether the navigation reflects what visitors actually need. This is process work that requires review, feedback, and iteration with a client who has real preferences and busy weeks.
Building the site in a maintainable way. The path from AI-generated design to a site a client can actually manage (in WordPress or a similar CMS) still has significant gaps in it. The build itself remains largely manual work.
Testing across devices, browsers, and screen sizes. A site needs to work reliably on phones from three years ago running outdated browsers. AI-generated designs frequently look great in one context and break in another.
Integrations. Forms that go to the right place. Analytics that track the right events. Payment systems configured correctly. Email marketing tools connected. CRMs synced. Every integration is small work individually and substantial work in aggregate.
Accessibility and performance. Meeting WCAG standards. Optimising for Core Web Vitals. These are technical disciplines that reward experience and attention. AI can help. It does not eliminate the work.
SEO fundamentals. Technical SEO setup, on-page optimisation, schema markup, page speed for search rankings, correct site structure for crawlers. Ongoing considerations rather than a one-time task.
Feedback loops with the client. Revisions, approvals, contract review, scope changes. These take the time they take because they involve decisions being made by busy people. AI does not accelerate the client’s calendar.
The compelling AI website demos share a common pattern. Someone types a prompt. An impressive-looking page appears. Nobody asks: what business is this for? Who is the audience? Where is the content coming from? What does success look like? Does the “About Us” page match the actual company? Do the testimonials belong to real customers? Is the pricing accurate? Does the site connect to the CRM, the email tool, the payment system?
The demos work because the content is placeholder and the constraints are minimal. In a real client engagement, the content is real and the constraints are numerous.
The other missing piece in most demos is what happens after the initial generation. The version that appears in the first minute usually needs several rounds of adjustment. Each round in a real project involves a client with an opinion, a calendar, and competing priorities.
The distortion this creates goes beyond client expectations. It is reshaping the market itself.
When clients arrive believing a website should cost a fraction of what it actually costs to build well, three things tend to happen. Some agencies compete on that lower price by cutting the parts of the work AI has not compressed: strategy, content preparation, testing, integrations, editability, accessibility. The site ships, looks polished, and starts failing quietly six months later when it needs to be updated, ranks poorly on search, or breaks on the audience the client actually has. Other agencies decline the work and lose it to whoever is willing to promise the timeline. A smaller number try to educate the client through the sales conversation and either succeed or lose the project to whichever competitor talked more confidently about their AI stack.
Realistic timelines for professional builds have not changed dramatically. Industry benchmarks put a simple brochure site at 1-4 weeks, a corporate or e-commerce site at 4-10 weeks, and a complex application at 3-6 months or more. AI-augmented workflows have trimmed the edges of these ranges without collapsing them. Agencies claiming otherwise are usually promising what they will not deliver, or are quietly stripping out the parts of the work the client cannot see until later.
The pattern is the same one now visible across other fields being reshaped by AI hype. The gap between what the technology can plausibly do and what its loudest promoters claim it can do gets exploited commercially. Whoever makes the boldest claims about AI-driven efficiency wins the sale. Whoever tries to describe the work honestly loses it. The market rewards the promise. The client discovers the difference after launch.
“AI-powered” has become a marketing category as much as a technical one. The same phrase can describe an agency using the tools well or an agency using them as cover for doing less work. AI itself has become essential to nearly every modern workflow. The distinction is in how it is being applied, and in what it is replacing.
For a business planning a website in 2026, the useful question is no longer whether an agency uses AI. Almost all of them do. The useful question is what specifically the AI is doing on the project, what still requires human judgement and effort, and what the agency is being paid for at each stage.
