I felt surprisingly nervous ahead of giving a careers talk to 160 A-level students at my old school last October.

My instinct was to share the key lessons learnt from an entrepreneurial career — scaling software businesses, building teams, managing risk. But I was acutely conscious that the world that today’s students are entering bears little resemblance to the one I experienced. My stories probably wouldn’t land. 

What might work better would be some timeless advice and why curiosity, agility and a diverse mix of skills are more essential than ever.

Enterprise newsletter

Advice and inspiration for entrepreneurs in The Times Enterprise Network’s briefing.

Sign up with one click

Like many of us do, I turned to AI for help to craft the speech. At first, I treated it as a clever editing and search tool — efficient, useful, but not transformative. Then came an unexpected twist.

Having co-drafted the talk, it suggested using an AI agent to create the slides. Within minutes, like Aladdin’s lamp, this virtual genie was sourcing images, deciding layouts and structuring my story faster than any designer I’d ever briefed. For the first time, it felt like genuine collaboration between human creativity and machine intelligence.

That small experiment opened a much bigger door. A few months later I had a personal task. This entailed all of the administration after a loved one passed away.

Again I turned to the AI for help. It responded: “Do you want to build an app to help you?” Coincidentally, I’d recently listened to a podcast on “vibe-coding” — plain-English prompting that generates functioning software. I decided to test the idea.

I gave the same brief to several of the leading AI tools and picked the one that felt most intuitive. In a flash, the genie was out of the bottle again — only this time, a few months later, it was older, wiser, and astonishingly capable.

As my prompts improved, the system began to anticipate what I wanted, even suggesting usability features and workflow improvements. Vibe-coding, I discovered, is a combination of creativity and precision. You learn to think like a product team: ask the agent to behave like a user interface designer for visuals, a data architect for structure, or a product manager for defining minimum viable product features.

I even asked it to suggest pricing models — I challenged its first recommendation and refined it until it matched both market logic and my experience.

Within a week, I had a credible prototype ready to test on real users that could scan images of items, suggest where to donate, sell or recycle them, and handle any payments. 

Vibe-coding hasn’t turned me into a software developer, but it’s certainly turned me into something new: an empowered problem-solver whose idea can be translated into a working product without waiting for a technical team. It’s like building an architect’s model before the building construction starts.

The more I experimented, the more I realised this isn’t just market “buzz”. Vibe-coding represents a deeper shift in how value is created in software. When AI can generate and test code faster and cheaper than ever, the defensible “moat” around businesses starts to move away from intellectual property and towards proprietary data, customer trust and scale.

We’re moving into an era of “compound engineering”, where AI systems don’t just help code, they embed across the entire development cycle: discovery, design, build, testing and runtime. Each cycle compounds productivity. That means turbo-charging early-career developers with AI tools, while the more experienced engineers are focusing on systems design, AI governance and scalability.

Naturally, this kind of change creates turbulence — and that’s showing up most visibly in the financial markets. As chair of La Fosse, the UK’s largest employee-owned tech talent business, I see first-hand how our clients’ needs are evolving. The conversation is no longer just about tech skills; it’s about adaptability, AI fluency and data integrity. We recently surveyed 2,020 UK tech professionals on their use of AI and the results highlight the maturity gap for the people at the vanguard of change.

While adoption is high, 35 per cent of workplace AI guidelines remain “limited” or “unclear”. Only 36 per cent of entry-level staff and 39 per cent of intermediate employees have had formal AI training. That highlights that governance and education aren’t keeping pace with enthusiasm, and wholescale workforce reskilling is now urgently needed.

Throughout my career, I’ve encouraged teams to “walk into the future” — to imagine how political, social, economic and technological forces will shape our world in five or ten years, and to take steps today to prepare for that future. AI agents and vibe-coding are tangible proof of how fast that future is arriving.

For me, these experiences haven’t been about learning to code. They’ve been about learning to think differently — to experiment, to test, to prototype, and to use technology as a creative partner rather than an obstacle.

That mindset is going to be vital for all of us, not just for coders and technologists but for anyone navigating a rapidly changing world.

As I told those students at my old school, the most valuable qualities you can cultivate have nothing to do with specific tools or job titles. They’re timeless: curiosity, flexibility, emotional intelligence and the courage to keep learning.

We often talk about the “future of work” as if it’s something distant, waiting to be discovered. But it’s already here, in every interaction between human and machine that helps us think or create in new ways.

AI isn’t just replacing us. It’s reshaping what we’re capable of. The sooner we learn to work with it — responsibly, creatively and with purpose — the better prepared we’ll be for what comes next.

Martin Leuw is chairman of La Fosse, a former chief executive of Iris Software and a part time vibe-coder