23+ Years in Web Development. And AI Is Changing How I Work.

I have been building websites and digital products since 2003. Over those years, the tools changed again and again — and I changed with them. But the current shift toward AI-assisted work feels different from the ones that came before.

For a long time, learning a new tool usually meant becoming faster or more capable at a particular part of the job. A new framework improved how I structured applications. Better development environments made everyday work smoother. Modern APIs opened the door to connected products.

AI is changing something broader: not just how quickly I can work, but how much of the journey from an idea to a working product I can meaningfully contribute to.

From websites to entire digital products

My first projects were much closer to traditional web design and frontend development. Over time, I moved deeper into PHP, application architecture, databases, integrations and backend development. Later, analytics, SEO, conversion tracking and digital marketing became increasingly connected to the systems I was building.

That progression was not a planned switch from one profession to another. It happened because real projects rarely fit neatly into one discipline. Building something is one challenge. Making sure it works reliably, can be found, can be measured and is useful to people is another.

I still enjoy writing code. But today I find the connections between those areas just as interesting as the code itself.

AI is an amplifier, not an autopilot

The most obvious benefit of AI is speed. It can help draft code, explain unfamiliar parts of a system, suggest tests, compare approaches and reduce repetitive work. That matters, but it is not the most interesting change for me.

The bigger shift is how naturally I can move between different kinds of work. In the same project, I may be thinking about a Laravel integration, then reviewing an interface, checking analytics data or shaping content for a marketing campaign. AI can support all of those conversations without requiring me to start from a blank page every time.

It makes experimentation easier. Questions that previously might have stayed on a to-do list can become a small prototype or a concrete comparison of options. The result is not always usable, but the feedback loop is shorter — and that changes how I approach problems.

AI changes not just the speed of development, but what one experienced developer can realistically achieve.

Experience matters more, not less

There is an important catch: AI can produce an answer that looks convincing and is still wrong.

It can miss a requirement, invent an API detail, introduce fragile assumptions or suggest an implementation that works in isolation but does not fit the existing system. It does not automatically know the constraints of a business, the expectations of users or the history behind a technical decision.

That is why I do not see prompting as a replacement for engineering judgment. The useful skill is not simply asking a tool to generate more output. It is knowing how to define the problem, provide context, challenge assumptions, validate the result and recognize when to say no.

After 23 years, I have seen enough real-world projects to know that a solution is more than something that runs on a developer's machine. Security, maintainability, performance, accessibility, privacy and operational realities still matter. None of those responsibilities disappear because AI helped write the first draft.

More room to connect development, data and marketing

Some of the work I find most rewarding sits between disciplines. A digital product needs thoughtful design and solid engineering, but it also needs to be understood and improved after launch.

My work on Cycletrack, for example, has involved much more than a codebase: product decisions, user experience, privacy considerations and the challenge of turning a personal idea into something others can use. Other projects have combined web platforms with data processing, analytics, SEO or lead generation.

AI makes it easier to explore those connections. I can review implementation details and then switch to interpreting a dashboard or outlining an experiment. It does not make me an instant expert in every field. It helps me collaborate with the tools, question the output and learn faster while relying on the fundamentals I have built over time.

Learning to work differently — again

I do not think experienced developers need to compete with AI at generating lines of code. What matters is the ability to turn unclear needs into good decisions and dependable outcomes.

That involves curiosity as much as experience. I still need to learn new tools, revisit old assumptions and accept that some familiar ways of working are changing. I also need to be honest about what AI does well and where it still needs close supervision.

Perhaps that is the surprising part. More than two decades into this career, I feel like I am learning how to work again.

And I genuinely enjoy that.

---

If you also work with AI, I would be interested to know: has it mainly changed how fast you work, or has it changed the kind of work you take on?