Skip to content

GETTING_STARTED.md

Latest commit

 

History

History
347 lines (241 loc) · 6.99 KB

GETTING_STARTED.md

File metadata and controls

347 lines (241 loc) · 6.99 KB

Getting Started with AI Automation Framework

This guide will help you get started with the AI Automation Framework from installation to building your first AI application.

Table of Contents

Installation

Prerequisites

  • Python 3.10 or higher
  • pip package manager
  • OpenAI API key (for cloud LLMs) or Ollama (for local LLMs)

Step 1: Clone the Repository

git clone https://github.com/yourusername/Automation_with_AI.git
cd Automation_with_AI

Step 2: Create Virtual Environment

# Create virtual environment
python -m venv venv

# Activate it
# On macOS/Linux:
source venv/bin/activate

# On Windows:
venv\Scripts\activate

Step 3: Install Dependencies

# Install all dependencies
pip install -r requirements.txt

# Or install in development mode
pip install -e .

Configuration

Set Up Environment Variables

# Copy the example environment file
cp .env.example .env

Edit .env and add your API keys:

# Required for OpenAI models
OPENAI_API_KEY=sk-your-key-here

# Optional for Claude models
ANTHROPIC_API_KEY=your-key-here

# Other optional settings
DEFAULT_MODEL=gpt-4o
LOG_LEVEL=INFO

Using Local LLMs (Optional)

If you want to use local LLMs with Ollama:

# Install Ollama from https://ollama.ai

# Pull a model
ollama pull llama2

# The framework will automatically detect Ollama on localhost:11434

Your First AI Application

Example 1: Simple Chat

Create a file my_first_app.py:

from ai_automation_framework.llm import OpenAIClient

# Create a client
client = OpenAIClient()

# Simple chat
response = client.simple_chat("Explain quantum computing in simple terms")
print(response)

Run it:

python my_first_app.py

Example 2: Chat with Memory

from ai_automation_framework.agents import BaseAgent

# Create an agent with memory
agent = BaseAgent(
    name="MyAssistant",
    system_message="You are a helpful programming assistant."
)

# Initialize the agent
agent.initialize()

# Have a conversation
response1 = agent.chat("What is Python?")
print("Response 1:", response1)

response2 = agent.chat("What are its main use cases?")
print("Response 2:", response2)

# The agent remembers context!
response3 = agent.chat("What did we just discuss?")
print("Response 3:", response3)

Example 3: Document Q&A with RAG

from ai_automation_framework.rag import Retriever
from ai_automation_framework.llm import OpenAIClient

# Create knowledge base
retriever = Retriever()

# Add documents
documents = [
    "Python is a high-level programming language known for its simplicity.",
    "JavaScript is primarily used for web development and runs in browsers.",
    "Rust is a systems programming language focused on safety and performance."
]

retriever.add_documents(documents)

# Query
query = "Tell me about Python"
context = retriever.get_context_string(query)

# Generate answer
client = OpenAIClient()
prompt = f"Context: {context}\n\nQuestion: {query}\n\nAnswer:"
answer = client.simple_chat(prompt)
print(answer)

Core Concepts

1. LLM Clients

LLM clients provide a unified interface to different AI providers:

from ai_automation_framework.llm import OpenAIClient, AnthropicClient, OllamaClient

# Choose your provider
client = OpenAIClient()  # For GPT models
# client = AnthropicClient()  # For Claude models
# client = OllamaClient(model="llama2")  # For local models

2. Agents

Agents combine LLMs with memory and capabilities:

from ai_automation_framework.agents import BaseAgent

agent = BaseAgent(
    name="MyAgent",
    system_message="Your role and personality here"
)

# Agents maintain conversation history
agent.chat("First message")
agent.chat("Second message - remembers first")

3. RAG (Retrieval-Augmented Generation)

RAG enhances LLM responses with relevant information:

from ai_automation_framework.rag import Retriever

# Create retriever
retriever = Retriever()

# Index your knowledge
retriever.add_documents(your_documents)

# Retrieve relevant context
context = retriever.get_context_string(query)

4. Tools

Tools extend agent capabilities:

from ai_automation_framework.tools.common_tools import (
    CalculatorTool,
    FileSystemTool,
    DateTimeTool
)

# Use tools directly
result = CalculatorTool.calculate("2 + 2 * 3")
print(result)  # {"result": 8, "success": True}

# Or register with agents for automatic use

5. Workflows

Automate multi-step processes:

from ai_automation_framework.workflows import Chain

def step1(input):
    # Process input
    return processed_data

def step2(data):
    # Further processing
    return final_result

# Create chain
chain = Chain(steps=[step1, step2])

# Run
result = chain.run(initial_input)

Next Steps

Learning Path

  1. Basics - Start with examples/level1_basics/

    • Simple chat interactions
    • Prompt engineering
    • Text processing
    • Streaming responses

  2. Intermediate - Move to examples/level2_intermediate/

    • RAG implementation
    • Function calling
    • Workflow automation
    • Document processing

  3. Advanced - Explore examples/level3_advanced/

    • Multi-agent systems
    • Complex task planning
    • Agent collaboration

Try the Demos

Run the interactive demos:

# Chatbot with memory
python examples/demos/chatbot_demo.py

# Document Q&A
python examples/demos/document_qa_demo.py

# Code assistant
python examples/demos/code_assistant_demo.py

Explore Features

  • Usage Tracking: Monitor costs and token usage
  • Caching: Speed up responses and reduce costs
  • Document Loaders: Process PDFs, Word docs, text files
  • Local LLMs: Run models privately with Ollama

Build Your Own

Start building your own applications:

# Your custom application
from ai_automation_framework.agents import BaseAgent
from ai_automation_framework.llm import OpenAIClient

# Create specialized agent
class MySpecialAgent(BaseAgent):
    def __init__(self):
        super().__init__(
            name="MyAgent",
            system_message="Your custom instructions"
        )

    def run(self, task):
        # Your custom logic
        return self.chat(task)

# Use it
agent = MySpecialAgent()
result = agent.run("Your task here")

Getting Help

  • Examples: Check examples/ directory for code samples
  • API Reference: See framework source code with detailed docstrings
  • Issues: Report bugs on GitHub Issues
  • Discussions: Ask questions on GitHub Discussions

Common Issues

Issue: API Key Not Found

Make sure you've set up your .env file with your API keys.

Issue: Module Import Errors

Ensure you've installed all dependencies and activated your virtual environment.

Issue: Ollama Connection Error

Make sure Ollama is running: ollama serve

Happy building! 🚀