AI Can Build a Website in a Weekend. Building a Good One Still Takes More Than That

AI can build a website in a weekend. We are seeing the proof of that across our industry. We are also seeing what happens next.

In July 2025, an AI coding tool deleted a live production database during an active code freeze, then generated thousands of fake records to cover its tracks. Recently, researchers found that 170 applications built on the Lovable AI platform shared the same critical vulnerability, exposing hundreds of API endpoints containing personal and payment data. The root cause was a single misconfiguration generated by default, creating a problem that was replicated across production websites at scale. Separate analysis by Aikido reported that more than two-thirds of organisations have found vulnerabilities in AI-generated code, even though around a quarter of all production code is now written by AI.

This is the gap nobody markets to you.

AI can ship a website that looks finished. Building one that holds up under real-world pressure still takes expertise, judgement and a plan for what happens after launch.

A few years ago, building a website still felt like specialist work. You needed designers, developers and strategists sitting around a table trying to work out how everything fit together. Even relatively straightforward websites involved planning, revisions and technical work before anything ever went live.

That process wasn’t always efficient, but it forced teams to think carefully about user experience, technical requirements and long term objectives before a website ever launched.

We’ve Seen This Pattern Before

The speed of change feels new. The pattern does not.

We saw this with drag and drop website builders.

Platforms such as Wix, Squarespace and Shopify lowered the barrier to entry and made launching a website far more accessible. More people could get online, test ideas and establish a digital presence without the same upfront investment that was once required.

That shift created opportunities, but it also introduced a new set of challenges. Websites launched quickly, only to become difficult to maintain as expectations grew.

Performance issues emerged. Integrations became harder to manage. Search visibility became more difficult to improve. Backend structures became increasingly restrictive as websites evolved beyond their original purpose.

Launching a website became easier, but building one that could scale alongside a growing organisation did not. We are seeing the next evolution of that trend.

AI generated websites feel like the next step in the same journey. The technology is different, but many of the lessons are familiar.

The Problem Usually Starts After Launch

We understand the appeal of AI generated development.

We use AI ourselves.

Used properly, it removes friction, accelerates prototyping and creates more time for solving meaningful problems. It has earned its place within modern digital workflows.

The problem is not AI, it’s assuming that speed removes complexity.

You can now generate layouts, write code, create imagery and build functioning websites from a handful of prompts. What previously took days or weeks can now happen in hours.

The process has changed significantly, but the fundamentals of building a successful website have not.

When websites are built rapidly through prompts, generated code and quick fixes, long term structure is usually the first thing sacrificed. Everything works on launch day. The pressure arrives later.

As traffic grows, new functionality gets added, integrations become more important and commercial demands increase.

That is when the cracks begin to appear.

A five-page startup site can become difficult to manage once ecommerce functionality, booking systems, CRM integrations and reporting platforms start getting layered in.

The website still works, but it simply becomes harder to scale, harder to maintain and harder to evolve.

Looking Finished and Being Built Properly Are Different Things

One of the reasons AI generated websites have attracted so much attention is that they can look convincing very quickly.

A homepage can appear polished, professional and launch-ready within minutes, but what you cannot see quite so easily is everything underneath.

The architecture, performance, accessibility, search engine foundations and long term maintainability that sit beneath the surface.

Those are rarely the things people notice first.

They are usually the things that determine whether a website performs six months, twelve months or three years later.

A website is not simply code and functionality. It is also how people experience your brand.

As more layouts, content and design decisions become automated, the value shifts elsewhere. Original thinking, creative judgement and an understanding of what makes people engage with a brand become increasingly important.

The web is becoming easier to populate, but standing out is becoming harder.

That is why experienced designers, developers and strategists still matter. Not because launching a website is difficult, but because building something sustainable requires more than getting online quickly.

Security Is Becoming a Bigger Conversation

Security is one of the least discussed parts of the AI website boom.

It is also one of the most important.

This is not theoretical for us. We have spoken with several contacts whose projects have been caught up in data breaches tied directly to AI-generated code. In one case, an AI-generated endpoint inside a booking application was hijacked and used to extort the business that deployed it. None of these incidents made the headlines. Most never do. They surface quietly, in conversations between people trying to work out how a system they thought was finished became the weakest link in their operation.

This is the part of the AI website conversation that rarely gets discussed publicly. The visible failures, like the Tea App breach, are useful because they make the risk concrete. Approximately 72,000 images were exposed, including 13,000 government ID photos used for user verification after attackers exploited weak authentication controls and API misconfigurations. Both are common signs of rushed development.

AI generated code is not automatically secure simply because it functions correctly.

It can still contain poor configuration choices, insecure defaults and vulnerabilities that only become visible when real users begin interacting with a system.

The data supports that concern. Research has found AI-generated code has 2.7 times higher vulnerability density than human-written code, while Veracode found AI tools failed to prevent cross-site scripting vulnerabilities in 86% of test cases.

The challenge is not only technical but commercial as well.

As Aikido CISO Mike Wilkes has highlighted, AI-generated development creates new challenges around accountability. Developers may not have written the code, security teams may not have reviewed it thoroughly and responsibility becomes much harder to define when something goes wrong.

Integrations Are Often Where Problems Start Appearing

If you run a modern website, it rarely exists in isolation.

Most websites sit at the centre of a much wider ecosystem, connecting CRMs, ecommerce platforms, stock management systems, reporting tools, customer portals and third party software. That is where complexity starts to build.

The booking application we mentioned earlier sat at exactly this kind of integration point. On the surface it appeared to be a straightforward customer-facing tool. Behind the scenes it connected multiple services, workflows and data sources, each relying on the others to function correctly.

That is often where problems emerge.

The challenge is rarely the feature people can see. It is everything happening behind it.

An online booking system might work perfectly on launch day. Six months later it could be connected to payment gateways, customer databases, reporting platforms, automated communications and internal business processes. What appears to be a single feature on a website can quickly become a critical operational system.

That is why integration planning matters.

Website integrations are not simply about connecting one platform to another. They are about understanding how information moves through an organisation, where dependencies exist and what happens when something changes.

This is one of the areas where AI-generated development can struggle. A prompt can build functionality, but it cannot always understand the wider business context that functionality needs to operate within.

The more important a website becomes to your organisation, the more important those foundations become.

What We Expect to See Next

The current AI website boom feels familiar because we have already seen versions of it before.

The pattern is familiar. Technology becomes more accessible, adoption accelerates and the benefits become obvious. Then the consequences begin to emerge.

Over the next few years, a growing amount of development work will involve untangling rushed infrastructure, improving performance and rebuilding websites that were originally designed around convenience rather than long term scalability.

That is not a criticism of AI, it is what happens whenever speed becomes the primary objective.

There is also a wider conversation beginning to emerge around the environmental impact of that rapid digital growth. As more websites, assets and infrastructure are generated at scale, sustainability is becoming a bigger part of the discussion. We explored some of those challenges in our recent article on AI and digital sustainability.

AI Is Changing Development. It Is Not Replacing Expertise

AI is now part of how every serious agency works, including ours. We use it to prototype faster, remove friction from repetitive work, and free our team to spend more time on the thinking that actually moves clients forward. It earns its place when used well.

But as the web fills with AI-generated content, AI-generated layouts and AI-generated code, the work that stands out shifts. It becomes the work that shows real thought, real judgement and real human creativity. The brands that win the next few years will pair the speed of AI with the depth of human expertise, not replace one with the other.

That is how we approach digital work. Insights that guide. Creativity that connects. Tech that performs. AI accelerates each of those. It does not replace any of them.