A modern programming language that makes developers happier.

kotlinlang

I'm happy to announce the first release of <https://github.com/mscheong01/interfAIce|interfAIce>, a new open source library written in Kotlin
interfAIce is an automatic proxy generation library that lets you harness the power of ChatGPT into your applications - just by defining an interface.
The auto-generated proxy will query the OpenAI api based on your method definition(name, parameters, returnType, etc) and return the LLM's response in any Kotlin/Java type you want!
The library also comes with support for:
• spring boot
• Kotlin Coroutines &amp; Java Reactor
If you're interested, check out my <https://medium.com/@mscheong01/harness-the-power-of-gpt-just-by-defining-an-interface-8ccfa0b94a8c|blog post >for more
Slack Conversation

I've read your blog post and readme and I still don't understand where would I use this? afaiu it creates implementation based on what AI returns? if so, will my application becomes undeterministic?

Well since the emergence of GPT4, many products that operate upon GPT results were released. This library just makes that kind of integration easier.

<@U2F6E34G1> Agreed, I’ve got about 10000 questions raised in my head. But I think the most obvious answer is that tests will become _even more_ important with this kind of technology. As long as your tests are fine, the implementation doesn’t really matter.

You're right. As for now, I don't see LLMs replacing actual code implementations. But there are still cases that only gpt could handle with its power to interpret natural language.
ex) you want to auto-classify user posts
```enum class PostType {
    GENERAL,
    QUESTION,
    DEBATE,
    ...
    ;
}
data class Post(
    ... // some Post class that you use in your application
)
interface PostClassificationService {
    fun classifyPost(post: Post): PostType
}
```
And in the future - who knows? maybe AI will evolve to become fast and precise enough to replace written implementations too