I vibe coded half a dozen apps. Here’s what I learned

In a world where artificial intelligence could fundamentally change the tech world in both exciting and terrifying ways, one particular AI use case has been gaining a lot of attention.

“Vibe coding” is the act of creating computer programs through AI prompting, using plain English, even without previous programming experience. It hasn’t made a big splash yet. As of August 2025, the use of vibe coding tools wasn’t widespread among the IT professionals I surveyed in the Spiceworks Community. Of ~50 IT respondents to my quick poll, fewer than 10% had used AI coding tools, and roughly half had never heard of vibe coding before.

Are these IT professionals missing out? Eager to test drive new AI functionality, I put Gemini’s Canvas tool to the test.

Can vibe coding make programming less painful?

Before we talk about the potential of vibe coding, a little background on myself: I earned a degree in computer science more than two decades ago. After coding and scripting professionally in various languages (Python, Bash, Visual Basic, etc.) for several years, I shifted to a more people-facing tech role. Since then, I have barely touched coding tools.

While I enjoyed creating with code — that feeling of getting something to work after hours of struggle was satisfying — I always thought the process of coding was much more tedious than it needed to be. Progress could be quite slow, and getting just one simple task to work could require a lot of trial and error, deciphering cryptic messages, and wasting time digging through arcane documentation.

In short, programming — a potentially fun and creative medium where you can create almost anything you can imagine — was much more frustrating than it had to be.

Vibe coding tools make programming much easier

After coding more than half a dozen apps using Google Gemini’s Canvas tool, I can confidently say that the app allows you to create, iterate, and even innovate very quickly.

In my first hour, I was able to build the basic framework for a few games … a platformer game where you have to collect objects, a crossword puzzle app, and a word scramble game. In truth, it was quite liberating. All I had to do was describe the game I wanted to play (i.e., A pong-like game where a human plays against a computer player) and the mostly rough shell of the app would appear in just one minute. From there, I could start getting creative instead of getting bogged down in the technical details.

Previously, getting to that point would have taken hours or even days. But with vibe coding,  anyone can instantly get a prototype up and running, leapfrogging many steps straight into the refinement or editing process. Dare I say, these tools made programming joyful again, at least for me.

Vibe coding tools aren’t perfect 

I was admittedly shocked at how much vibe coding simplified the task of creating apps. Using AI to create applications removes many of those barriers to entry. However, if you’re expecting a tool like Google’s Gemini to spit out code on the first try, think again.

You can’t expect to get a perfect result with just a single prompt.  With any type of coding, there are going to be some bugs and glitches. To get the apps to where I wanted them to be, there was still a lot of testing and telling the AI which minor bugs to fix. I also had to clearly specify which UI or visual elements to add or change.

You’re also very likely to run intocoding dead ends via prompting. A couple of times, no matter how many times I told the app what I wanted it just couldn’t deliver due to fundamental logic errors. At that point, I had to dig into the code a little bit to see where the errors were occurring, and then suggest a different approach.

There are also technical limitations. For example, it’s very easy to write simple client-side applications that don’t require interactions with existing services. But once you start wanting to save information on a server somewhere, you’re going to need to have your own infrastructure, and an understanding of how those interactions work.

Communication: An essential vibe coding skill

In a recent interview with the CTO of Spiceworks’ parent company, Ziff Davis, Joey Fortuna, he stated: “The new programming language is pure English, and AI is becoming increasingly accessible and easy to deploy.”

I agree 100%. Vibe coding is opening doors to more people. And ironically, to be effective at AI vibe coding, human communication skills are key. Like in conversations with real people, being able to clearly articulate what you want, often stating your requests in very clear and unambiguous terms, or in multiple different ways (using examples) can go a long way into getting what you want from AI.

According to Fortuna: “If you’re reasonably smart, you’re reasonably creative, you have reasonably good communication skills, and you’ve had a couple of years of professional experience, and you have an eye for the right AI tool set, you can really succeed.” Importantly, you need to have a clear vision for what you want your app to do and be able to articulate that vision to the AI.

Who benefits from vibe coding?

Currently, I think vibe coding tools are most beneficial to those who already know how to program, at least a little, because you will have to debug and refine your apps. That said, as long as you have a fundamental understanding of computer logic (not necessarily the nitty-gritty coding syntax), you can go far with vibe coding tools.

No longer do you have to spend years learning even the ins and outs of a programming language, or struggle with debugging obscure issues. Vibe coding tools make it relatively trivial to create basic applications, and you can refine these apps through simple English prompts, without getting bogged down by some of the more mundane technical aspects of coding.

Humans still needed in a vibe coding world

People are still integral in a world where vibe coding makes coding easier. An LLM won’t magically understand all of the dependencies that exist in your environment or the needs of your organization, nor will it understand how to defend against vulnerabilities in your specific situation.

Importantly, vibe coding doesn’t necessarily yield secure, optimized code that’s ready to deploy in a production environment. Vibe coding is really best for prototyping and exploring possibilities.  Someone who knows what they’re doing should be in the loop at all times, and any code should be reviewed thoroughly before it’s deployed and made available for wider use.

Additionally, when everyone has the power to write programs, organizations need to pay extra attention to AI security and governance. Like with Shadow IT and shadow AI, “shadow vibe coding” might become a bigger concern.

In the end, vibe coding is really just a tool, but a powerful one at that. To answer the question posed in this article, I do believe IT professionals are missing out if they don’t gain an understanding of vibe coding.

In my opinion, the mantra, “always keep learning,” applies perfectly to this situation.

To stay relevant in a world that’s increasingly dependent on AI, IT professionals need to understand emerging tools… not only to leverage them in their work, but also to support and manage them, as well offering strategic advice on how to best utilize them.