Emergent Labs’ Mukund Jha discusses vibe coding and how AI is changing the software development landscape.
We’re officially in the age of AI-built software. The excitement around vibe coding has skyrocketed since Andrej Karpathy popularised the term in February last year. Now, rapidly maturing AI-powered tools are letting anyone, regardless of coding skill, take software from idea to production-ready reality.
It’s easy to get caught up in the hype when an estimated 41pc of all global code is now AI-generated. Much of the discussion around vibe coding has been centred on speed and cost. If anyone can create production-ready software, why would you pay anyone just to develop software?
There’s no question that vibe coding will change how software is developed, but the real story here isn’t about building apps faster or cheaper. Vibe coding is completely restructuring the sources of software value.
It’s a new era, one where creativity, curation, and continuous evolution become the real differentiators, not just how much code your LLM spits out divided by cost.
‘Human taste is now one of the most important tools in a software engineer’s arsenal’
Value in the age of software abundance
While the ability to generate code quickly without resorting to hiring expensive developers is undeniably part of the technology’s appeal, this perspective fails to see the bigger picture.
We’re about to enter a post-scarcity era of software. AI is already significantly reducing the cost of designing, building and testing new software.
In a world where any company (or individual) can spin up production-ready applications quickly and cheaply, the competitive edge will shift from how software is created to how well it integrates into user workflows, supports teams and drives business outcomes.
When software becomes abundant, the value shifts from creation to integration. Winners will be defined by how well their products align with workflows and deliver results, not by the volume of code they generate.
As a result, the software economy will start to look very different in terms of what is sold, where and how.
In a post-vibe-coding world, composability will take precedence over complexity as users move away from monolithic tools toward modular, lightweight agents that plug into existing workflows and evolve organically. Curation will cut through the clutter as AI tools flood the market and the value of structured marketplaces and internal catalogues that rank tools by accuracy, ROI and latency increases.
Perhaps most importantly, outcomes will come to matter a lot more than ownership as licensing models shift from seat-based pricing to usage and delivered value, especially in enterprise environments.
Domain knowledge over coding skills
Vibe coding is seeing enthusiastic adoption from individual entrepreneurs and small business owners all the way to the Fortune 500, where 87pc of companies now report using it in some capacity.
“This isn’t a fad. This isn’t going away,” Garry Tan, president and CEO at Y Combinator, said on an episode of the Lightcone podcast. “[Vibe coding] is now the dominant way to code and, if you’re not doing it, you might just be left behind.”
Tan’s attitude isn’t surprising given the enthusiastic adoption vibe coding has seen within Y Combinator itself. Among Y Combinator’s Winter 2025 cohort, 25pc of start-ups reported having codebases that were 95pc AI-generated.
Well-funded, cutting-edge start-ups with access to some of the world’s best coding talent are still heavily buying into vibe coding. Is it any wonder that some people say we’re looking at the end of software engineering as we know it?
The reality is that vibe coding doesn’t mean the end of software engineering. However, it will change how software engineers create value.
Because the value of vibe coding lies outside cost and speed, the technology really shines in its ability to move developers closer to the problems they’re trying to solve.
Say your sales department is having difficulties with client onboarding. Traditionally, sales would have to work with a software engineer, describing their needs and working with the dev team to build and integrate a solution.
With vibe coding, someone from that sales team who is as close as possible to the problem (and therefore best positioned to understand the solution) can build the right tool for the job. As a result, domain knowledge is becoming more valuable than pure coding skill. Software engineers are becoming product managers.
Meet the citizen developer
At the same time, a whole generation of non-developers who were previously held back by the gap between their ideas and their coding knowledge are about to enter the software economy. Individual entrepreneurs and small business owners will be able to quickly craft software to support their ambitions, rather than hiring expensive developers or buying cumbersome integrations off the rack.
For example, agentic vibe coding platforms are giving individuals the opportunity to solve their thorniest problems without massive investment or teaching themselves to code.
One user in Germany built a comprehensive AI-powered social media and marketing operating system to automate their entire workflow from brand analysis to content creation, entirely built to suit their needs.
In London, a marketing consultant with no coding background built five applications, including a voice-enabled appointment reminder SaaS to better organise and structure his time and support his workflow.
It’s not just individuals and small business owners who will increasingly take ownership over software development. Every company will be made up of ‘citizen developers’. These people could be anyone in the organisation who is close to a particular problem and, with the help of vibe coding, can tackle those problems more quickly and efficiently, putting their deep domain knowledge to practical use.
Vibe check: what’s next?
Vibe coding is breaking down the barriers between the highly technical roles that execute a designer or product manager’s vision and the vision itself. Human taste is now one of the most important tools in a software engineer’s arsenal.
If AI tools make it possible for anyone to make a functional, professional-grade application, the key differentiator is no longer the ability to write good or even great code. Instead, we’re headed for a future where the ability to envision new applications, craft UX and put a personal stamp on software that will make some applications shine brighter than others.
By Mukund Jha
Mukund Jha is a co-founder of Emergent Labs, a company that builds autonomous coding agents and has been backed by Y Combinator and Lightspeed Ventures.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
