We’ve all been there. The clock is ticking on the launch date; the client’s expectations are sky-high. In the past, this would have signalled the kind of pizza-fuelled, sleepless nights of high-pressure coding and debugging that developers like me (occasionally) thrive upon.

But things are changing. Coding tools with AI features are helping us to do far more with less, adding a thrilling new layer to what digital teams can achieve.

For a recent high-stakes build we took on, the only way we could realistically deliver our vision on time was to lean on AI-assisted coding tools. In fact, doing so was about ten times faster than coding by hand.

It made me reflect not only on how AI tools can accelerate delivery – but also how they’re reshaping what it means to be a digital developer today.

‍Stepping up to the challenge

‍The brief was simple: get the build live in time for a major global event. But the timeline wasn’t. With so many unknowns and so little time, a traditional development process just wouldn’t have cut it.

It seemed appropriate to embrace AI for the client in question, a leading AI company with a big vision for the future. And crucially, it never felt like cutting corners or outsourcing our skills; the goal was to amplify them to bring our ambition to life.

Quick, keen and able to crunch through requests faster than our small team could alone, the tools acted like an extra junior developer in the room. Without them, we wouldn’t have hit the deadline.

‍One giant leap for coding

As an agency, we’ve been using AI to amplify our creative and strategic work for some time. But as little as a year ago, I wouldn’t have seriously contemplated leaning on AI for such a high-stakes client project. The tools existed, but the results were clumsy; they often took longer to fix than to write from scratch. More like the work of a wayward intern than a buttoned-up junior developer.

Things have progressed at an astonishing pace. Today’s tools integrate directly into our coding environment. I can “chat” with them, ask them to implement features in specific files, and accept or reject changes line by line. Inline suggestions anticipate what I’m trying to do with uncanny accuracy. It’s not always perfect, but it’s impressively productive.

‍The quality quandary

‍On this project, I could tick off tasks in minutes that might have taken me hours alone. The AI’s ability to reference vast libraries and documentation saved huge amounts of research time.

But speed came with a cost: quality. The initial codebase was bloated, repetitive and not easily maintainable. AI tends to break the DRY principle (“don’t repeat yourself”), duplicating code rather than structuring it for reuse.

This created technical debt: code that works now, but slows things down in the long term. For short-lived projects, this tradeoff may not matter. For anything more permanent, it really does. Impressive as the tools now are, the lack of sophistication proves that they still need expert human oversight to be truly useful.

‍Changing roles‍

Using AI is changing the way we work as developers. Instead of spending hours hand-coding, my job is increasingly to set tasks precisely and clearly, review outputs critically, and guide the AI towards the best practices I’ve learned through experience. I’m more like an architect, orchestrating the build rather than laying every brick myself.

That’s both exciting and unsettling. It requires judgment, problem-solving and oversight. But it also moves developers further away from the satisfaction of writing code that got many of us here in the first place. We might even miss those Deliveroo-fuelled late nights.

While AI can accelerate delivery, it can’t yet replace the human ability to imagine, design and structure systems thoughtfully. We need to keep hiring, training and nurturing our living, breathing junior developers.

I would advise new graduates to learn to work with AI and treat it as an essential tool. But don’t skip the fundamentals: if you don’t understand how code works at a basic level, you risk becoming locked into AI’s decisions, unable to adapt or maintain what it produces. Clear, well-structured code is still valuable, and still very human.

‍Creativity matters

‍As all developers know, coding is more than just typing instructions into a machine. It’s a creative process. The logic, structure and elegance of code are based on choices that can’t be outsourced to an algorithm.

AI is perfect for projects with immovable deadlines, prototypes and quick iterations. It’s less ideal for long-term builds where clean, maintainable code is crucial.

Because AI doesn’t imagine; it just assembles patterns. It can generate functional code, but it won’t dream up a radical new solution. It doesn’t care about whether the end product is elegant, lightweight or satisfying. That’s our job.

Ultimately, speed may win the deadline, but creativity and craftsmanship win the long game.