Yes, you need to code your UI component yourself. And Prompt Enginner yourself to have your custom fn to be called.

@@kale_bhai oh I thought It was creating components on demand as v0 is capable of doing so, but yeah that'd be really slow. so it's just function calling for a RSC with the data props as whatever the LLM came up with as info ?

That's right! You provide it with a bunch of functions and components, and it will call the appropriate function and render a component when the conversation calls for it.

Not necessarily, you build the components upfront. But the LLM decides when to render them and what data to pass as props. So the components are not necessarily generated by AI. They can be though, if you give the LLM access to an eval() function and let it spit out JavaScript strings.

rookie numbers

We're giving frontend developers more tools to be productive. v0 doesn't replace building UI for example, but it helps my get a decent first version I can build for. I still need to hook up all the business logic and polish things. It's an accelerant for building great products :)

Why are you so negative 🤔

​​@@leerobVercel is awesome, I like this AI, making building decent UI way much easier ❤

Front end dev is the horse staring at the car.

I have no idea whats in your mind but I am very excited about Vercel AI, it gives me the capability of building sleek UI easily, this is really awesome ❤

Why are you this negative? I feel very excited because from now on I am able to deliver decent UI even though I am not a UI/UX designer, this is sweet ❤

No. This is basically building different type of user interfaces. Instead of coding the UI to use navigation button, we basically code the UI to interact with the AI. We make premade components (duty of frontend engineer) and link it with AI, code it so that the AI detects the persons need from the message they send to the AI ("Please show me the weather in Dubai"), it will detect and tell your code that weather function is called, your code handles the logic based on the parameters AI sent you (detects country=dubai), then your code will send back the data to AI and then AI will decide to stream the premade component you programmed with the params it detected from user's message. Thats all, so AI just decides when to render component, it doesn't actually generate the UI automatically, thats what you have to do aka the front-end engineer. (This answer is just the jist of it so it might not have perfect details but overall the flow is correct)

What in the world are you talking about? 😅😮😢

@@maskman4821 If they keep on making and improving these AI stuff what in the world will happen with us

But I really appreciate the harkwork of these people 🙂

@@user-cn3cu4go3sThe same thing that happened when tractors were introduced into farming, and all the farm workers thought they would lose their jobs… there was an agricultural productivity boom and more farm workers were needed. The same thing that happened when desktop computers were introduced and all the admin staff thought they would lose their jobs… there was a productivity boom and more staff were needed to build, implement, train, use and repair the computers. There’s an expression that I’ve heard a few times over the past months, “You won’t be replaced by AI; you’ll be replaced by someone who knows how to use AI”!

Why are you scared? You should be excited about this AI, making building decent UI way much easier is awesome ❤