My coding skills leave much to be desired. I never used tutorials and guides long enough to actually create the apps and programs I wanted to see. AI chatbots based on large language models have changed this. Now I can write down my vision for something like a horror movie visual gallery or a variation of a classic computer game, and the chatbot will turn my simple English into hundreds of lines of code—and bring my vision to life.
It's called vibration coding. For experienced programmers, this can speed things up significantly. But for non-programmers, it puts programming tools in our hands. Having the right mindset during this process can be the difference between a good or bad experience, even if you don't achieve the results you originally hoped for. Sometimes you'll need to learn what you can and can't do as you code.
You can choose Gemini, ChatGPT, Claude or another chatbot and each will have its own features and quirks. Your approach and how you prompt the chatbot is as important as the model you choose, since everything is based on your instructions.
Here are five tips that can help you get started with Vibe coding.
Know your chatbot
The chatbot you use matters, although the type of model may matter more than who created it.
When trying create the same application Having used both Gemini 2.5 Flash and Gemini 3 Pro, it became clear that a different approach was needed with the former to achieve a similar result to the more advanced Pro model. This means more specific intentions and guided instructions and, in my experience, more manual work.
When I coded with Gemini 2.5 Flash, it often made the process more hands-on, which was not what I was looking for. When refining the application, it would only provide the code for the section that was changed, and I would have to replace the code manually or use a different prompt to provide all the code. On the contrary, when using Gemini 3 Pro, it will automatically provide all the code.
Your experience may vary, but if you are choosing between using a “fast” or a “thinking” model for your vibration coding project, you may need more skill and training to work with the fast model. If all you have is a fast or less advanced model, and it asks you to make changes to the code that you are not comfortable with, you can again ask the chatbot to provide all the code (or set a rule to provide all the code with changes after each change is made).
From my experience using several (but not all) chatbots, if you don't mind getting your hands dirty with a little code, you'll likely have success with most models if you provide clear and specific prompts. Non-programmers: stick to more advanced models of “reasoning” or “thinking” if you can. I've found that thinking models do the heavy lifting for you.
Getting to know your chatbot, its limitations and capabilities is based on experience, so it's best to use it in general and ask questions.
Make your prompts specific
If you have a detailed idea of which project's code you want to use, specificity is key.
Coding Vibe starts with what's in your head. If you have a very clear idea, you can and should include everything you want to see in your project in the initial invitation. Make it comprehensive. The chatbot will create whatever it can and you will hopefully see your idea take shape.
On the other hand, you could simply have a very general idea of what you want the application to be and give the AI more control over how the application functions or how it achieves what you asked it to do in a different way.
Iteration, iteration, iteration
Your dream app probably won't make it into the first or second invitation you give your chatbot. You'll likely spend more time clarifying than anything else.
Ask your chatbot for suggestions. If you are stuck on any part of the web application, be it its functioning or its style, just ask the chatbot. He will provide you with as many offers as you want. And since Vibe coding is an iterative process, asking for five ways to improve your app can really make a difference.
Get a little technical
Non-programmers will have to understand some technical things to make the application or project work well. The good thing is that you can use a chatbot as a resource.
Here are some things you should find out:
Format: For most simple web applications, HTML output is ideal for non-coders. And if you don't want to deal with multiple files and folders, you can request a single HTML page. This can lead to limitations and potential “memory issues” in the chatbot's context window if the file becomes huge, but it is one of the easiest ways to get the output code into your browser for testing. If you're not sure, ask the chatbot which format is best for your project and app launch process.
Scale: If you are not sure about the capabilities of a chatbot, just ask it. It will tell you that it won't be able to create a new social network for you, but if you want an audio visualizer that replicates the old days of Winamp, you're in luck. If your request is beyond the chatbot's capabilities, ask it to suggest alternative approaches.
Check for errors: Sometimes a chatbot will provide broken code, so you will have to test it again and again. It's even more important to communicate what's not working or exactly how you want something to work. If you provide a vague hint, don't be surprised if you get a vague result. Telling a chatbot “this doesn’t work” is significantly less effective than detailing the specific errors you encountered during testing. The more specific your feedback, the more accurate the AI's response, making the overall experience more efficient and enjoyable.
Be open-minded
Unless you have a very specific and rigid goal for your vibration coding project, it is important to keep an open mind. If you don't know anything about coding, you probably don't know all the options that are at your fingertips. If you're unsure of your limitations, it's just as easy to underestimate what you can do as it is to overestimate.
In an ideal world, all your code will run smoothly, but errors are likely. Whether it's a chatbot's memory forgetting something, a technical limitation, or something in between, some things can break. If you're lucky, the chatbot will recognize and fix the problem, but sometimes it won't be able to, and that's where it helps to have an open mind. Ask for alternative approaches – you may find an approach you like better than your original idea. You will never know what you don't know unless you ask.
Don't be afraid to start over
As with other creative processes, sometimes starting over is for the best. If you've gone through countless iterations and your app is still nowhere near what you want it to be, consider starting over. This can be done entirely from scratch, or you can take the code from the first chat and use it as the basis for a new conversation you start.
While you can instruct your chatbot to start from scratch on the original chat, moving forward may require a clean slate. Try again in a new chat so the AI doesn't get confused about your previous project. You may have identified some cues that could have set your project back that you can avoid this time, allowing you to focus on what worked rather than keeping what didn't in the chatbot's memory for reference.
A fresh start is needed not only to ensure that the chatbot “gets it right” this time, but also to help you reboot your creative flow.
