Get Started with AI (mostly GenAI)

How do I get started with GenAI?

Like anything else, the best way to get started is to start. However, here are some helpful tips to guide you.

Think of a use case
1. Sounds simple, but the first thing you have to do is figure out what problem you're trying to solve
2. Does the problem even need GenAI or can it benefit from GenAI?

Various levels of complexity

The skill level needed for any specific task varies depending on how you're trying to use GenAI
If you're thinking of a flow where you want to use a normal GenAI response from any model based on an input, that's the easiest flow.

sequenceDiagram
     participant System
     participant User
     participant Assistant
     User->>Assistant: Hey!
     Assistant->>User: How are you?

The next level would be adding context to provide information outside of the LLM.

sequenceDiagram
     participant System
     participant User
     participant Assistant
     System->>Assistant: Context: 
     User->>Assistant: Hey!
     Assistant->>User: How are you Sky?

This is how RAG (Retrieval-Augmented Generation) was invented. The context you provide can be as simple as a name or as complex as a dictionary of books and authors. 4. Now let's add system prompt and some more context.

SystemPrompt: "You're an assistant who helps user find their favorite books. Here are list of books and authors: "
UserPrompt: "Can you help me find books by Kafka?"
Assistant: "Yes, I can help you find books by Kafka. Here are some books by Kafka: The Metamorphosis, The Trial, and The Castle."

sequenceDiagram
    participant User
    participant System
    participant Assistant
    System->>Assistant: SystemPrompt: "You're an assistant who helps users find their favorite books..."
    System->>Assistant: Context: 
    User->>Assistant: Can you help me find books by Kafka?
    Assistant->>User: Yes, I can help you find books by Kafka. Here are some books by Kafka: The Metamorphosis, The Trial, and The Castle.