“Cleave” means to split apart and to cling together. The same word. Opposite meanings. Both valid.

Two Old English words sounded similar enough that everyone just stopped distinguishing them, and the language said fine, one word can mean both things.

What’s in store:

Read Time: 5 minutes

AI TOOLS

A beginner’s guide to Hugging Face

If ChatGPT, Claude and Gemini are the AI tools you use, Hugging Face is where you go to see what else is out there.

Yes, the name is a bit… odd.

Hugging Face is a huge platform where developers, researchers and companies share AI models, datasets and demos. People often describe it as the GitHub of AI.

But why would the average vibe coder care?

Imagine you tell Claude:

“I want my app to remove the background from uploaded photos.”

Claude can help you build the feature. But you may still need an AI model that actually does the background removal.

That’s where Hugging Face comes in.

1. Find AI models

Head to Models and you can search for AI built to do specific jobs, including:

Think: “I need AI that does X. Does a model already exist?”

Click a model and you’ll usually see its Model Card — basically its information page.

It tells you what the model does, who made it, how to use it and, importantly, what license it has.

Pro-tip: If the page looks like gibberish, paste the link into Claude and ask: “Explain this model to me like I’m a beginner. Could I use it in my project?”

2. Try AI before using it

If Models feels too technical, start with Spaces.

Spaces are AI apps and demos you can try directly in your browser.

You’ll find tools for things like:

You’ll also see some common suggestions at the top. 

All of this is especially useful when vibe coding.

Before asking Claude to add a background-removal model to your app, for example, you can test a few options and see which one actually works best.

Try this: Search Spaces for something like background removal or audio transcription and test a few results.

3. Compare models

You’ll also find leaderboards and benchmark results comparing different AI models.

You do not need to become an AI benchmark expert, but this is where you want to go to see whether a model is actually useful. It’s kind of like checking reviews. 

Especially if you hear about it somewhere else, search for it on Hugging Face, check its Model Card and see how it performs.

Then ask Claude to explain the numbers if necessary. But, at the end of the day, you may need to try it yourself. 

4. Peek at AI datasets

Hugging Face also has a Datasets section.

Datasets are basically collections of text, images, audio and other information used to train or test AI.

You probably won’t spend much time here as a casual user.

But if you’ve ever wondered:

“What are these AI models actually learning from?”,  this is where you can peek behind the curtain.

5. Run AI on your own computer

Some Hugging Face models can be downloaded and run locally, meaning on your own computer instead of somebody else’s servers.

Why bother?

It can mean more privacy, more control and potentially fewer API costs.

The catch: some models require a powerful computer. Apps like Ollama and LM Studio are making local AI easier, but you absolutely do not need to mess with this when you’re starting.

Do you need Hugging Face Pro?

Probably not.

If you’re just browsing models, trying Spaces and figuring out what Hugging Face even is, stick with the free account.

The paid plan makes more sense once you’re actually building with Hugging Face regularly.

Where should you start?

Start with Spaces.

Search for an AI task you already understand and try a few demos.

Then search for the same thing under Models and open a Model Card.

That’s enough.

You don’t need to learn Python or understand the entire Hugging Face ecosystem.

The main thing to remember is: Claude can help you build the app. Hugging Face can help you find the AI that powers part of it.

Screenshot this

If your vibe-coding tool keeps botching one stubborn part of the build- and you have no desire to learn actual code - Models can help fill the gap before you spend 40 prompts begging Claude to fix it. - HL

You can get our free bundle of 250 prompts to make money below!

THAT’S ACTUALLY WILD

In late 2024, cocoa prices surged to around $13,000 per tonne, smashing a record that had stood since 1977. Bad weather and crop diseases helped drive the spike, pushing chocolate makers to shrink bars, raise prices and experiment with cocoa-free alternatives. Prices have since fallen nearly 70% from their peak, and some manufacturers plan to bring back their original recipes (please hurry up with this).

Tool of the Week

AirMusic

AirMusic generates original songs, AI singing voices, and covers from text prompts, with voice cloning, stem separation, and music video creation all built in for royalty-free content you can use anywhere.

Check it out! →

Mystery Link

We don’t know where it goes either…