I have been building things digitally for most of my adult life. It started with a Sinclair ZX Spectrum that my father bought as a single Christmas gift for the whole family back in 1984 or 1985 — one home computer instead of individual gifts, which felt incredibly baffling at the time but turned out to be one of the most consequential decisions of my childhood. My brother and I spent weekends poring over BASIC manuals, building games from scratch, restarting every time Kenya Power knocked out the electricity and wiped out our meticuously created BASIC code. At some point I remember typing into that machine in plain English — asking it to just make me Pacman, a popular 80s arcade game — and getting an error back. It spoke BASIC. I spoke the language of hope. These two things are not logically compatible! 

From there the journey ran through a TIME Magazine cover in February 1996 that pointed me towards the Internet as a career — Marc Andreessen, barefoot on a golden throne, co-founder of Netscape, the story titled “The Golden Geeks” — through a first internet job at Form-Net, building websites on Notepad, a personal site on Geocities, and eventually an Aptech certificate in 1999 where I got my first serious coding exposure to Visual Basic. This tool mattered because it was, at the time, the first genuinely accessible way to build software applications — Object Oriented, event-driven, drag-and-drop interfaces on forms, code attached to button clicks. You did not need to be a computer scientist to build something functional. It was revolutionary. But it was also just the first chapter of a much longer story.

Every decade since has added another chapter. Dreamweaver. WordPress. No-code tools. Low-code platforms. Each one promised to lower the barrier further. Each one did — but only partially. There was always still a gap. Always still a threshold that kept most people on the consumer side of the digital economy rather than the creator side. I founded Dotsavvy in 2002 and have spent over two decades watching that gap frustrate brilliant people with brilliant ideas who simply could not bridge the distance between what they imagined, and what they could actually build.

That gap has now been closed. Completely. The name for what closed it is vibe coding.

This Is The Biggest Shift In The History Of All Things Digitally Created

Let me be direct about what has actually happened here, because I think it is still not landing with the weight it deserves.

ChatGPT launched in November 2022 and hit 100 million users in two months — the fastest digital consumer platform adoption ever recorded to-date. That was the starting gun. Since then we have moved through Generative AI, which gave us the ability to create content and code from prompts, into Reasoning Models, which gave AI the ability to genuinely think through complex multi-step problems, and now into Agentic AI — where AI systems do not just respond to a prompt and wait, but take a goal, plan a sequence of actions, use tools, iterate, self-correct, and execute entire workflows end-to-end with minimal human intervention.

When you apply Agentic AI to the act of building digital products, you get vibe coding. Vibe coding does something that no previous wave of tools has managed: it collapses the gap between having an idea and being able to build it — not for a niche group of technically inclined people, but for anyone. A student. A market trader. A doctor. A farmer. Even Mama Mboga. A first-time entrepreneur with a problem that nobody else has solved yet. Anyone.

The phrase was coined by Andrej Karpathy earlier in 2025 and it captures the experience exactly. You are not writing code. You are directing. Describing. Iterating. Think of it like being the director of a film — the vision is entirely yours, and the crew, in this case the AI agent(s), handles every technical detail of bringing it to life.

What You Can Actually Build — & The Scale Of That Possibility

This is where it gets genuinely exciting. The range of what is now buildable through vibe coding is extraordinary — and deliberately limitless at both ends of the ambition spectrum.

At one end, you can build something just for yourself. A personalised budgeting tool that works exactly the way your brain works. A custom dashboard that pulls together every metric you actually care about from your business. A meal planner that knows your dietary preferences and your weekly schedule. A project tracker built precisely around how your team operates. Tools like these — hyper-personalised, built for an audience of one — have never been possible without a developer. Now they are an afternoon’s work.

At the other end, the same tools and the same approach are being used right now to build what could become the next generation of technology companies. Startups are going from idea to functional product in days rather than months. Founders are testing market hypotheses with working prototypes before they have hired a single engineer. The cost of building a minimum viable product (MVP) has dropped so dramatically that the barrier to starting a technology business is no longer capital or technical skills — it is the quality of the idea and the commitment of the person behind it.

The platforms driving all of this are moving fast and each brings something distinct to the table. Claude Code and Claude Cowork from Anthropic bring full agentic development and task automation into your workflow — the agent reasons through architecture, navigates codebases, and executes changes autonomously. Google has entered the arena hard with Stitch, which generates polished, production-ready user interfaces straight from prompts, and Google AI Studio for building complex agentic applications on top of Gemini. ChatGPT Codex from OpenAI, Replit, Cursor, and Lovable each offer their own take on the same fundamental shift — describe what you want to build, and watch an agent build it.

Taken together, these platforms represent something genuinely unprecedented: a complete, end-to-end stack for digital creation that anyone can use meaningfully, from day one, without any prior technical background.

I Know This Because I’ve Been Vibe Coding Too

Over the last two weekends I built two complete digital platforms — end-to-end — using vibe coding tools and agentic AI prompting. No manual code. I described. I directed. I reviewed outputs and iterated. What came out, across hours rather than months, were functional, designed, and essentially launch-ready products.

I will be honest with you about the experience, because I think honesty here matters more than hype. It was not always smooth. Managing context across a long build session takes discipline — the agent can lose the thread of earlier decisions if you are not careful. Communicating complex debugging issues requires precision even if you are not writing the fix yourself. Getting the architecture right early saves a lot of pain later. These are real challenges and anyone serious about vibe coding needs to go in with their eyes wide open.

But here’s the thing. None of those challenges are technical barriers in the traditional sense. They are about clarity of thinking, quality of communication, and progressive learning. Which brings me to the only honest caveat in this entire conversation.

The One Thing You Actually Have To Do

Vibe coding is not a magic wand. It is a profound amplifier of human intent — and like any amplifier, it magnifies both clarity and confusion in equal measure. The tools are accessible from day one, yes. But proficiency comes with practice. The more time you invest in understanding how these tools work, how to structure your prompts, how to manage context and iteration, how to think architecturally even without writing architecture yourself — the more powerful what you build becomes.

The barrier to entry in the world of vibe coding is no longer technical. It’s commitment and curiosity. You have to be willing to put in the hours, consume the learning, iterate through the frustrations, and progressively build the kind of fluency that turns a promising idea into something that genuinely works and genuinely adds value — whether to your own life or to the world.

That is a very different ask from what it used to take to build software. But it is still an ask. The people who answer it are going to build some remarkable things. Mark my words. 

What This Means For Kenya & Africa, Right Now

I think about this a lot from an African context, because the implications here are not abstract — they are immediate and they are enormous.

The digital economy conversation in Kenya and across the African continent has always been framed around access: to infrastructure, to capital, to technical skills. Vibe coding does not solve the infrastructure question overnight. But it does something unprecedented to the skills and capital questions simultaneously. A young person in Kisumu or Mombasa or Kampala or Nairobi or Dar es Salaam, with a smartphone, a solid internet connection, a clear idea, and the willingness to put in the learning time — can now build something that competes on a global stage. All this without a development team, significant upfront capital, or years of computer science training.

That young person could be building a tool just for themselves that makes their daily life better. Or they could be building the foundation of the next African technology unicorn. In 2026, both of those outcomes sit behind the same door. The key to that door is not a computer science degree. It is a prompt, an agent, and the commitment to keep going until what you have built is genuinely extraordinary.

My 12-year-old self typed into a ZX Spectrum in plain English and asked it to make him a game. He got an error. The gap between what he imagined and what the machine could do for him was, at that moment, essentially infinite. That gap is now gone. For everyone.