Simple Agentic Chatbot (Groq + Llama 3.1 8B)

A minimal Agentic AI framework built for learning and experimentation.

This project demonstrates how a Large Language Model (LLM) can decide:

• ➜✅ Whether to answer a query directly

• ➜🔧 Or call a tool when needed

• ➜🔁 And return back to the LLM after tool execution

It is designed for understanding the fundamentals of Agentic Systems.

State Graph is

    +---------+
    | __start__ |
    +---------+
          |
          v
 +------------------+
 | tool_calling_llm |
 +------------------+
    |            |
    |            v
    |        +-------+
    |        | tools |
    |        +-------+
    |            |
    +------------+
          |
          v
    +-------+
    | __end__ |
    +-------+

Mermaid Diagram

flowchart TD A [start] --> B[tool_calling_llm] B --> C[tools] C --> B B --> D[end]

Architecture Flow

start → tool_calling_llm → (tools if needed) → tool_calling_llm → end

Flow Explanation

Flow Explanation

1. User sends a query

2. LLM (Llama-3.1-8b-instant via Groq) decides:

   • If tool is required → move to tool node
   • If not required → respond directly

3. If tool is called:

   • Tool executes
   • Output is sent back to LLM

4. LLM generates final response

5. End state reached

📌 Why This Project?

• This project is intentionally simple to help you understand:

• What is an Agent?

• What is Tool Calling?

• How LLMs decide actions?

• Basic Agentic Architecture

• State Graph Workflow