Vibecoding: Promise, Risks, and What Comes Next

Software development has been changing a lot recently. Just as low-code and no-code website building platforms changed who could have a website, a new concept of vibecoding is reshaping how the code itself is created, and who can be called a coder. Instead of focusing traditional development workflows, vibecoding is about writing code with AI agents through conversational commands.

The result is a style of creation where even non-developers can build functional applications and websites. Vibecoding reflects a shift toward easier software creation.

Let’s look how vibecoding (code-generating agents) is different from AI website builders, how it impacts hosting market, and what turn of events can we expect next.

What is Vibecoding?

The term vibecoding describes a style of programming in which developers use conversational prompts to generate and refine software with AI tools. Instead of manually writing every function or configuration file, users describe the intended outcome and collaborate with an AI system that produces working code.

"Vibe-coding" was first coined in the early 2025 by Andrej Karpathy, the former director at OpenAI.

The phrase gained prominence in 2025 and was named Collins Dictionary’s Word of the Year. Collins defines vibecoding as the use of artificial intelligence to translate natural-language instructions into code, reflecting a broader shift towards conversational development workflows.

This approach builds on earlier advances such as autocomplete tools and code generators but extends them into full conversational environments capable of generating applications and services from prompts.

Website Builders and Vibecoding

In our articles, we have already talked about AI website builders, tools that implement AI in the core of the editor.

Vibecoding emerged a bit earlier than AI website builders, in the beginning of 2025. In essence, it shares similarities with AI-assisted development platforms:

Vibecoding generally has a broader use. AI website builders, as the name suggests, are focused on creating and editing websites. Their main audience is entrepreneurs and non-devs who are focused on their business and don’t have resources to spend on creating a website from scratch. Vibecoding, on the other hand, is more useful to developers, since it allows them to write code not just for websites, but also for apps. It is suited for rapid prototyping and experimentation.

Possibilities and Drawbacks

Conversational coding is new, so it holds a lot of promise. However, this is not a magic pill that turns anyone into a senior programmer.

Advantages

There is documented evidence that AI-assisted coding can increase productivity and accessibility when used with appropriate oversight. A report on Anthropic’s Claude Code found that AI coding tools were able to handle complex engineering tasks and, in some cases, significantly increased developer productivity.

Similarly, participants in an AI-assisted coding hackathon — including teenagers with limited formal training — were able to build functional applications within 24 hours using conversational coding tools.

Drawbacks

Despite its advantages, vibecoding also presents well-documented risks when used without careful review.

AI-generated code can contain more defects and vulnerabilities than human-written code. The analysis from Techradar reported that AI-generated pull requests contained significantly more issues and security risks than those written by developers.

There are also examples of misuse. Investigations into phishing campaigns found that attackers used AI site-generation tools to create convincing fake websites quickly, demonstrating how conversational development tools can lower barriers for cybercrime.

Vibecoding and Hosting

As AI-assisted development becomes more common, the relationship between software creation and hosting infrastructure is also changing. AI-generated applications involve continuous updates, real-time processing, and dynamic scaling requirements that differ from traditional static websites.

Hosting providers have begun adapting infrastructure specifically for AI-assisted development workflows. One example is Vietnamese VinaHost, which has introduced a concept described as Hosting Vibe Coding. The service is designed to support AI-driven development environments and was developed in response to changes in how code is generated and deployed.

The company describes its approach as focused on flexible infrastructure capable of scaling with AI-generated workloads, maintaining responsive connections to AI services, and adapting security measures as code changes dynamically. While this model reflects emerging infrastructure strategies for AI-assisted development, its long-term adoption across the hosting industry remains uncertain.

Deployment of vibecoded projects on Vinahost. Source: Vinahost

Final Advice

Right now vibecoding is a new tool in developers’ kit, but knowing how interfaces are changing with AI implementation, it’s hard to fully predict what to expect next.
However, here are a couple of tips for hosting companies and programmers themselves in regard to the use of vibecoding.

For hosting companies

Prioritise compatibility with AI‑generated, composable stacks. Vibe‑coded projects often use modern frameworks and AI‑generated multi‑file structures. Hosting platforms need to support Node.js, React, Vue, Vite, serverless functions, and Git‑based deployments to handle these fluid, rapidly‑changing codebases.

Vibe coding emphasises speed over architecture, so hosting must be stable even when the generated code isn’t.

For developers

Always review and test AI‑generated code. Multiple guides stress that vibe coding works only when humans stay in the loop.

You should review generated changes, test functionality, document decisions, and avoid shipping unreviewed prototypes to production.

Skipping validation leads to higher rates of security flaws and missing validation logic, especially when developers rely on AI for CRUD scaffolds or automation.