Some people are saying that "vibe coders can't create anything new".
But this is completely incorrect.
Using just one Novel AI feature in an app can make a new experience for users.
The best way to do this is to leverage one of the latest and best APIs paired with your deep understanding of a problem people have (Ideally that you have).
Let’s start with APIs:
An API is a simple connector that lets your app borrow someone else’s tech.
For example, adding one API key can drop OpenAI chat, ElevenLabs voice, or Runway video straight into your vibe-coded project.
That means users could come to the app you build with other people’s tech and pay YOU MONEY.
And to set this up, all you need to do is tell the vibe-coding tool’s agent which API you want to use, and then insert your API key.
These are my favorite API’s to use in my vibe coded apps:
Each of these API’s give an experience to the user, and all of them have an input and an output.
As you likely know OpenAI just released their new image model.
You can use their new image API, which gives you access to use their tech in your tool, and you could create an instagram story generator on many vibe code tools in a matter of minutes with no coding experience. All you would need is an OpenAI API key.
Learn more use this prompt in Perplexity, ChatGPT, Claude, or Grok:
How do I get my OpenAI API key from OpenAI and test their API’s in the playground.
Let’s look at another example, AI video generators:
On the Runway website you can use their video generator simply by typing your idea for a video.
Text in → Video out.
But most of the people using Runway ML are generating videos for a specific reason, as part of a specific workflow.
For example they are likely copying and pasting all of these videos into premiere pro in order to edit a youtube video.
And since Runway has millions of users they can’t cater their product to every use case.
That’s where you come in.
You could vibe code a Premiere Pro extension that analyzes your video and generates B-Roll videos automatically and imports them into your timeline in a consistent style.
And since it’s so easy to create something like this, you may not even want to monetize something like this. You may just want to use it yourself or give access to your team.
Which is great. Using API’s allows you to become more efficient in your workflows that can be improved by linking these technologies together.
If you create an app and charge for it, the math is simple:
Estimate average cost per user. GPT-4o runs roughly $0.05 to generate a 400-word reply; ElevenLabs hovers around $0.30 per thousand characters; Estimate / observe how many requests they make per month.
Triple it for margin and bursts. If a typical user costs you $0.60 a month, a $3 “Pro” tier leaves plenty of cushion.
The bigger the problem you are solving, the more you will be able to charge.
Getting Started Creating an App that uses an API
Ok so you want to create an app that uses an API to power its core feature. You have two options:
Option 1 – Cursor
In order to build a custom chat app with GPT-4o you can use Cursor.
Using Cursor, first open the editor and create a new project.
Tell Cursor, “Create a chat app that lets me chat with an AI model about _____,” and it will generate the code for you.
Next, visit the OpenAI dashboard at https://platform.openai.com/account/api-keys, click Create new secret key, and copy the string it gives you. Create a file named .env in the project root and paste the key inside like this:
OPENAI_API_KEY=sk-your-key-here.This will allow you to securely create an app that uses the OpenAI API.
Option 2 – VibeCode. In order to reach the same goal even faster, open the VibeCode app and ask the agent, “Build me a chat app that uses GPT-4o in the style of:”
VibeCode handles everything so you don’t even need to create an API key.
