I founded INDIGO Concept more than 30 years ago and have spent the years since designing, building and managing websites for businesses of all shapes and sizes.
During that time, I have seen plenty of technologies arrive with the promise of making website creation faster, cheaper and easier.
Some have improved the way we work. Others have changed the industry completely. AI may prove to be the most significant of them all.
But after three decades of seeing what happens to websites not just at launch, but months and years later, I believe there is an important distinction being overlooked.
"Generating a website is not the same as knowing how to design, build and maintain one properly."
Website technology has always evolved
When we first started building websites at INDIGO, the process was very different.
Sites were smaller, browsers were inconsistent and much of the work involved solving technical problems that most users never saw.
Since then, the industry has moved through several major changes, from content management systems and responsive design to mobile-first thinking and increasingly sophisticated integrations.
Each development has made certain parts of the job easier. That is a good thing.
I do not believe complexity should be protected for the sake of it, or that businesses should continue paying for work that can now be done more efficiently.
At INDIGO, we use modern tools, automation and AI where they genuinely improve the work.
But every technological shift has also brought the same misconception: that making a website easier to produce somehow removes the need for experience. It does not.
The tools may change, but the important questions remain much the same.
- What is the website there to achieve?
- Who is it for?
- How will it need to change as the business develops?
A website can look finished long before it is ready
The appeal of AI website builders is easy to understand. A business owner can describe what they want and, within a short space of time, have something that resembles a complete website. It may include professional-looking page layouts, imagery, navigation, forms and copy.
On first inspection, it can be impressive.
The difficulty is that most business owners can only judge what they can see.
- Does the site look good?
- Does it work on their phone?
- Does the contact form send an email?
These are all reasonable questions, but they reveal only part of the picture.
A website can look perfectly respectable while carrying problems that remain hidden until the business tries to rely on it.
That may happen when a new service needs to be added, an integration stops working, or the business needs to bring in a developer to make changes or extend it. You may even find that the website is no longer being indexed by search engines, or that it is not accessible to users with certain disabilities.
In my experience, the real quality of a website is often revealed much later than launch.
Experience is not simply access to information
AI has access to an enormous amount of information. It can produce layouts, write code, generate content and suggest solutions within seconds.
What it does not have is lived experience. It has not spent years dealing with the consequences of earlier decisions. It has not seen how a shortcut at the beginning of a project can create a much larger problem later.
It has not dealt with a website failing during an important campaign or a previously reliable piece of software becoming unsupported.
Experience is not just knowing how something can be done. It is knowing which approach is appropriate, what should be questioned, where the risks sit and which compromises are sensible. It is also knowing when not to build something.
That judgement is difficult to see in the finished design, but it is often where the real value lies.
The problems usually appear later
An AI-built website may work well at first. The issues often emerge gradually. Pages generated in isolation can create small but noticeable inconsistencies in layout, messaging and behaviour. A site may also perform well in basic testing, but become less reliable as it evolves or is used in different circumstances.
Performance, accessibility and code quality can all appear “good enough” on the surface while still creating limitations underneath.
None of these issues necessarily makes the website unusable. But together, they can make it fragile, restrictive and expensive to improve.
AI is not the problem
It would be easy for someone in my position to argue that AI should not be used to build websites. That is not my view.
AI is already a valuable part of website design and development. It can help explore ideas, accelerate repetitive tasks and assist with code and content. We use it at INDIGO, and I expect its role within our work to grow.
The problem arises when generation is mistaken for:
- Producing a layout is not the same as understanding whether it suits the business.
- Generating code is not the same as knowing whether it is robust and maintainable.
- Writing content is not the same as understanding the audience or identifying what really needs to be said.
A tool can provide an answer. It cannot take responsibility for whether that answer was right.
Responsibility matters
One of the less obvious benefits of working with an experienced agency is accountability. When we design and build a website, we are not simply producing a set of pages. We are making decisions on the client’s behalf, asking questions they may not yet know to ask and considering how the website will be managed once it is live.
We are also thinking about whether the decisions being made today will still make sense as the business changes. And when something goes wrong, we are there to investigate it, explain it and put it right. AI does not remain accountable.
It does not retain the reasoning behind a decision or manage the relationship between the website, the business and the people using it. It does not carry the commercial consequences when its recommendation turns out to be wrong.
This does not mean that AI is useless. It means it needs to sit within a properly managed process.
Ownership is often overlooked
One of the first things to establish with any AI-built website is who actually controls it.
The business may own the domain, but have limited control over the website itself. It may be tied to a particular platform, subscription or account, with no straightforward way to move the site or access the underlying code.
These restrictions can seem unimportant while the website is being created. They become much more significant when the business wants to grow, change supplier or add something the platform does not support.
For a small or temporary website, that may be an acceptable trade-off. For a site responsible for generating business or supporting important day-to-day activity, it needs more careful consideration.
Does an AI-built website need replacing?
Not necessarily. A complete rebuild should not be the automatic recommendation.
Some AI-built websites are fundamentally sound and need only professional refinement. Others may have a good visual direction but require technical improvements beneath the surface.
There will also be sites that are so restrictive or fragile that starting again is the more sensible commercial decision.
The important thing is to understand which situation you are dealing with before investing more time and money. A proper review should establish what is working, where the risks sit and what can realistically be improved.
How we would assess it at INDIGO
At INDIGO, we would look beyond the visible design and try to understand the website as a whole.
That starts with ownership and access. We need to know what the business controls, how the site is hosted and whether it can be backed up, moved or worked on by someone else.
We would then review how well the site has been put together. Does it behave consistently across different devices? Does it reflect the brand properly? Are there technical weaknesses likely to affect performance, accessibility or future development?
