Why Vibe-Coding Raises Enterprise Governance Concerns
Software development is undergoing a transformation as natural language interfaces allow non-programmers to generate working code through conversational prompts.
This shift has prompted both enthusiasm about democratising development and sharp concern about code quality at enterprise scale.
Collins Dictionary has named vibe-coding its Word of the Year for 2025, capturing a moment where AI-assisted software development has moved from experimental to mainstream.
The term, coined by OpenAI co-founder Andrej Karpathy in February 2025, describes using AI prompted by natural language to write computer code.
The practice allows individuals with limited programming knowledge to create functional applications using tools including Replit, Cursor, Google AI Studio and Microsoft’s GitHub Copilot.
Rather than learning syntax and debugging logic, developers can describe what they want in plain language and watch the code materialise.
Andrej introduced the concept on social media platform X, stating: “There’s a new kind of coding I call ‘vibe coding’, where you fully give in to the vibes, embrace exponentials and forget that the code even exists.”
Why Google reports 30% of new code is AI-generated
The adoption rate reveals how quickly this approach has gained traction.
Sundar Pichai, CEO at Google, reveals that over 30% of new code developed at the search and cloud computing company now comes from AI.
That’s a substantial portion of the codebase at one of the world’s largest tech firms.
Research by Fastly, a content delivery network provider, also found that over half of code released by a third of senior software developers was AI-generated.
However, there’s a telling split in how different experience levels approach the technology.
Junior developers show less adoption, with only around 13% shipping code where more than half was developed by AI.
The reason? Senior developers with over 10 years of experience can spot and fix mistakes efficiently, making AI tools a genuine productivity boost.
Junior developers often struggle because they lack the pattern recognition to quickly resolve errors in AI-generated code, turning what should be a time-saver into a debugging nightmare.
Andrej notes that LLMs are “getting so good” but acknowledges these limitations candidly, adding: “Sometimes the LLMs can’t fix a bug, so I just work around it or ask for random changes until it goes away. It’s not too bad for throwaway weekend projects, but still quite amusing.
“I’m building a project or web app, but it’s not really coding - I just see stuff, say stuff, run stuff and copy paste stuff and it mostly works.”
The warnings against enterprise adoption without strategy
Raymond Kok, CEO at Mendix, a Siemens-owned low-code development platform, takes a harder line on vibe-coding’s readiness for enterprise environments.
While the approach enables rapid prototyping, the code produced often lacks the architectural structure required for business-critical systems.
Bug-prone outputs create real headaches when organisations attempt to integrate AI-generated applications into existing enterprise infrastructure.
Raymond says: “Businesses must not get ahead of themselves and blindly follow the vibe coding trend without full consideration of how it will affect wider IT strategy.
“Organisations must take time to build a comprehensive strategy fit for the long haul. When setting out governance plans, they should keep adaptability front of mind, to ensure they are able to leverage the best AI has to offer now and are ready to implement novel technologies yet to come.
“But adaptability should fit in with the governance fundamentals: organising your data, updating your SDLC to include data engineers and end users and investing in talented employees who are eager to evolve.”
Yet the challenges run deeper than debugging.
Enterprise software must meet non-functional requirements including security standards, performance benchmarks and compliance with industry regulations.
AI-generated code often focuses purely on functional requirements without addressing these concerns.
Raymond concludes: “Vibe coding in the world of enterprise software will need to understand the notion of non-functional software requirements, governance and control.
“Until then, it’s a short-term con with limited long-term gains.”
