Coding. Being 100% humanitarian, I always thought of it as some sort of sorcery. Just think, people write strings of random symbols that turn into an app people use everyday.
The downside? Coding is very hard and boring as hell.
And still, I do envy people who know how to code, and always wanted to try it myself. But as it was really tough for me, I've never gone through the basics. So I thought hey, it just happens, not your thing, move on.
That's when the modern go AI-yourself era presented me one of its gifts:
A new word, how lovely. But the deal itself is very promising: you don't have to spend half a year learning how to code – AI will do it for you. Does it come true in practice? Not really, and let me tell you why.
Join me on my vibe coding Odyssey, where I tried to vibe code, fell in love with it, made my first project, failed miserably on the way but didn't give up on it. I will also tell you what worked for me and how you can make your vibe coding experience less painful.
TL;DR
→ Vibe coding is a great thing if you’re interested in developing something simple
→ For content marketing, vibe coding works magic if you want to create something interactive and engage with your audience
→ You can code a rapid prototype or a basic MVP using AI tools like Lovable, but it can take quite some time
→ Vibe coding is not a dev replacement. Expect bugs, retries, and careful prompting
→ Rules that matter: be ultra-specific; iterate ruthlessly; force the tool to explain each step.
So what is this vibe coding thing?
On the surface, the concept of vibe coding couldn't be simpler. You have an idea and an AI coding assistant by the hand. You type in your prompt, the assistant does some thinking and builds you a ready-made product. If you ever dealt with low-code/no-code platforms, it's quite similar, but instead of drag-and-dropping ready-made templates, you command the AI assistant to code everything. That’s why they added “vibe”, I guess.
And here comes the however. Just like low-code/no-code platforms that are very easy to use and let you make cool stuff with their help, AI coding assistants are very limited in terms of their functionality.
AI-assisted development and copilots have been there for a while, and even developers are using them, but only to speed things up. AI tools make mistakes, a lot of mistakes. To give you an idea, the winner of a recent AI coding challenge managed to solve an astounding 7.5% of the coding questions published on GitHub.
So yeah, if you are about to launch the next Binance or Microsoft thanks to vibe coding, please don't. As of now, vibe coding is very good for rapid prototyping or basic MVP development, and this is where we draw the line. Remember, 7.5%, not 75%.
Some insights from my vibe coding experience
Now comes the practical part. Being an editor, I'm not looking to change my profile in the near future. But what's really interested me in vibe coding is the chance to create something that would blend in with the content my team and myself are producing.
That said, here’s how I created a project from start to finish using Lovable, one of the popular tools for vibe coding.
Act I. Entering the vibe coding abyss
Let's set the scene. My goal was quite simple: create a digital template library to display the examples of content my team has created, and let users copy their favourite pieces. For developers, this doesn't sound really intense, but for me, it felt like conquering Everest.
To clarify, here is how I compare to a typical developer.
Given the circumstances, the only viable option was to use AI tools like Lovable. There are alternatives like Claude Code or similar, but after some research of Reddit threads, I learned two things:
1. Beginners like myself only use Claude for developing calendar apps.
2. Claude requires you to double-check the code it generates, which is a big no-no for me.
Lovable, on the other hand, acts on its own: you just chat with it and it does all the heavy lifting for you. That’s what they say right away on their website.
My enthusiasm was through the roof, so I signed up and started exploring the platform.
Act II. First steps, no tests, just vibes
Once I registered, I had a clear idea in my mind and how to implement it. Since I've been working with AI tools like ChatGPT for a while (and even wrote an article on how to train it), I knew that such tools are all about being specific. So I entered my first prompt, and it worked!
Just one prompt and the site is ready. To say I was in awe was an understatement. It’s like you lived in a cave clashing stones to get fire, and then suddenly find yourself in the Bladerunner setting. Just mind blowing.
And there’s more to it. Lovable shows you the code it created to make your wish come true.
The process went on, and 10 prompts in, the first template library version was ready.
That's when Lovable offered me to perform code refactoring as the code it wrote became quite lengthy. What a lovely (pun intended) and clever idea, I thought to myself. And that's where things went south.
Act III. The crashdown and rubber duck debugging
Code refactoring was supposed to bring more stable performance, which was a huge plus for an expanding template library. Sounds very harmless, so I asked the assistant to perform it.
The good news is, the code refactoring has been performed. The bad news? It took half of the library as the sacrificial lamb.
A complete disaster. Instead of 46 templates I was left with 13, all the categories got renamed, the descriptions got rewritten and didn't match the templates. This was a classical occasion of AI hallucination, but compared to when ChatGPT starts babbling nonsense, this hallucination was very catastrophic.
So what should you do in this situation? You need to come back to square one and start debugging. But the problem is that I didn't know where the bug was. So if I wanted to find and eliminate the bugs, I had to make the AI tool itself apply rubberducking.
The thing is that rubber duck debugging works great if you're a real human being and say things out loud to something inanimate. But if you are an AI tool and have to do it yourself? Well, that's tricky.
Luckily, Lovable always shares with you what it does, so asking it to perform debugging while ALWAYS telling what it does specifically and why was a viable strategy. Step by step, we were moving towards a working project.
The result? The library was restored. Yes, it took me about 20 prompts to add more templates and perform code refactoring that wouldn't crash the whole project again, but hey, it worked.
Act IV. The finish line
To cut the long story short, after the hallucination incident got sorted out, the process was more than smooth. New templates were added, along with new categories. We even added a custom watermark (yes, copyright everything is the rule of thumb).
The final result was a working template library, ready for any further expansion. 74 templates split into 12 categories, a custom font, integrated watermarks, a popping contact us button – that vibe coding project definitely took off.
And some final numbers. To develop a template library, which includes 74 templates, a custom watermark functionality, a built-in contact us form and a CTA banner, it took:
- ~2,500 lines of code
- ~42 prompts
- ~6 work hours
Of course, for developers, it's quite a small project, but for a content editor, it's like building a super app. There were some ups and downs, but the result was well worth it.
If you want to learn more on how you can use Lovable to create projects like this, we made a dedicated post on LinkedIn.
Now, let's sum things up.
Are vibe coding tools actually helpful?
Oh yes, they are. But you need to curb your expectations. If you are a CEO of a software development firm, you can't use Lovable instead of hiring new developers. But if you’re looking to develop a simple solution, vibe coding tools like Lovable are a great fit.
Overall, opt for vibe coding tools if you want to:
- Enrich your content with interactive elements (quizzes, clickable templates, or a booking one-page website)
- Create a webinar event
- Build a prototype to hand over to real developers
And what does vibe coding mean for content marketing?
Well, a lot, to be honest. In the end, content writing is a craft, and you need to always master new techniques and approaches for your content to remain fresh and engaging. We've got just 8 seconds to have someone's attention, so we need to be smart.
Will content marketing change with all these innovations? Definitely, and people even suggest new roles in content marketing. Take Jimmy Daly, who already distinguished multiple sub-categories.
Personally, I don't expect such a division. The reason is somewhat straightforward: if you don't want to consider technological advancements and make something better, you're going to be left behind. For example, are you using landline phones as much as your smartphone? Same goes for vibe coding tools – if you can use it to make your content stand out, then better do it.
Our team can attest to this. For example, Lesya, our writer, has recently created a small quiz for her article about LinkedIn personas. And the best part? It took her just one prompt.
Kate, our CEO, also used Lovable and created a website for businesses to start their product-led content transformation journey.
All in all, vibe coding isn’t just another gag, but one of the things that came with the AI shift and is here to stay. So try it out, make your own mistakes, but don't give up.
Just remember, to succeed with vibe coding tools, you have to follow two main rules:
- Be as specific as possible.
- Be patient, because it won't work on the first try.
And one last thing before you go. If you like the idea of vibe coding projects blending into your content, we know how to get it done. So send us a note and let's zmistify your content.
