Vibe Coding: Taking New Ideas, From New People, Seriously

For decades, the path to technological power required learning a foreign tongue: Python, C++, or perhaps JavaScript. Today, the most powerful tool is plain language. You don’t program the machine: you talk to it; you vibe with it.

If we think back to Kenya to the early 2000s, the explosion of cheap mobile phones didn’t just let people chat; it allowed a farmer to text a market town, learn the latest price of maize, and strengthen his bargaining power.

We are now at a similar frontier with the advent of vibe coding, which allows anybody to give plain-language instructions that are turned into technical commands and scripts, or code. Vibe coding enables the people who actually experience a problem to build the tool to fix it. We are moving from mass-produced software to ‘artisanal’ software, where the barrier to entry has shifted from the everyday software engineer to the ordinary citizen. 

Just as cell phones sped up access to valuable information in developing countries through voice calls or basic texting, this kind of agentic coding may open comparable opportunities for whole new groups of individuals. By “talking” to an AI agent, an individual with an idea can discuss a problem they’ve identified, go back and forth to refine a solution, and create technical tools to put their innovation into action. That means anybody has a shot at building groundbreaking tech innovations.

Historically, writing software has required a lot of highly technical coding skills and math/computer science background. It has been dominated by a fairly narrow demographic — often highly educated men, typically white or Asian, and concentrated in specific regions like Silicon Valley or Bangalore.

Now, you don’t need to know Python or JavaScript to create valuable digital products. People with strong domain knowledge but without much technical background are empowered to build new things. 

Call it citizen software. Anyone, anywhere in the world — from teachers, students, and parents to community organizers — can define problems based on their own lives — and build tools to address them.

The apps we’re all familiar with, including Uber, Airbnb, dating apps, and social media, turn relationships and experiences into transactions and metrics. Their success is a tribute to their convenience, but they’re driven by a narrow ecosystem of elite engineers, founders, and investors. Yet many important problems — loneliness, neurodivergence, local educational needs — aren’t obviously profitable and don’t fit neatly into this model.

What happens when more people outside the world’s elites can build tools to address some of these pressing concerns? Imagine education tools like a bilingual homework explainer for a specific immigrant community, a dyslexia-friendly study planner designed by students themselves. Or in the expanding world of workforce education and training, informal or gig workers could use their phones to keep track of micro-credentials they’ve earned to qualify for certain jobs.

In the Global South, citizen-led coding could enable thousands of locally driven solutions to local problems, instead of one-size-fits-all corporate platforms. Many historical examples, from the printing press to the Kodak Brownie to cheap online blogging, show that lowering previous barriers can have a transformational impact: ordinary people can produce and share ideas and innovations rather than just being consumers.

But the promise isn’t always easy to realize. That’s certainly the case with agentic coding and AI. A big, accessible marketplace of apps will inevitably have huge variations in quality. Many apps are likely to be clunky or ineffective. In education, the stakes are high and many guardrails to ensure quality-control will be essential.

Democratically created apps will often have tiny audiences. Influential institutions and companies are inevitably going to keep controlling scale and staying power. The privilege associated with existing advantages, including education and networks, not to mention capital, massive data centers, and foundational AI models, mean that the Global North will still have a huge head start.

So empowering more individuals by lowering tech barriers is just one step in a much bigger project of reducing inequality. Vibe coding represents a profound shift in the technology class system, but it’s not sufficient for social and economic progress. The technology elites still own the infrastructure of the AI ecosystem: the platforms, the AI models, the data centers. 

Still, we have good reason to be cautiously optimistic about the implications of the radically accessible coding era. This doesn’t mean every social problem will be solved with an app. But it does mean the assumptions and values of traditional tech engineers and venture capitalists in Silicon Valley need not dominate the landscape and crowd out other ideas.

People can increasingly use tech to directly address what matters to them most. In education, that allows teachers, students, and lifelong learners to move beyond being passive adopters of edtech. They can become co-creators of their own educational pathways, including personalized degrees.

Just as the spread of cell phones showed that your access to information could unleash entrepreneurial energy, vibe coding may show that when you lower the barriers to building software, you unleash hidden reservoirs of imagination about what education and technology are for.

Whether that promise is realized will depend less on the power of AI itself than on who gets invited to use it — and on whether we are willing to take everyday ideas from everyday people seriously.