AI Research Planner

AI-powered research planning and execution system that creates intelligent research plans and gathers comprehensive data from the internet.

Features

• 🧠 AI-Powered Planning: Generate intelligent research plans using configurable AI models
• 🌐 Internet Research: Execute plans by searching and scraping web data
• 🧹 Smart Data Cleaning: Clean data intelligently, skipping cleaning if it removes too much content
• ⚙️ Multi-Model Support: Works with Ollama, OpenAI, Anthropic, and other AI providers
• 📊 Progress Monitoring: Track execution with detailed feedback and logging
• 🎯 Complexity Levels: Simple, standard, deep, and comprehensive research modes

Installation

!pip install ai-research-planner
OR
pip install ai-research-planner
CLI Use:
ai-research research "Latest AI developments 2025" --config config/config_openai.yaml --store-var my_results --verbose
OR
research-planner research "Latest AI developments 2025" --config config/config_openai.yaml --store-var my_results --verbose

# ========================================
# GOOGLE COLAB SETUP AND USAGE EXAMPLES
# ========================================

# 1. Install the package in Colab
!pip install ai-research-planner

# 2. Set up your OpenAI API key (recommended way)
import os
os.environ['OPENAI_API_KEY'] = 'your-openai-api-key-here'

# 3. Import and use the research planner
import asyncio
from ai_research_planner.main import ResearchPlanner

# ========================================
# METHOD 1: Direct Assignment with = sign
# ========================================

async def research_with_assignment():
    """Research and store results with = assignment."""
    
    # Create planner
    planner = ResearchPlanner()
    
    # Configure for OpenAI
    planner.config.set('ai_model.provider', 'openai')
    planner.config.set('ai_model.model_name', 'gpt-4o-mini')
    
    # Research and assign to variable with =
    my_results = await planner.research_to_variable(
        "Latest AI developments in 2025", 
        complexity="standard"
    )
    
    return my_results

# Run in Colab cell
my_results = await research_with_assignment()

# Now you can use my_results directly
print(f"📊 Total items collected: {len(my_results['raw_data'])}")
print(f"🧹 Cleaned items: {len(my_results['cleaned_data'])}")

# Access specific data
urls = [item['url'] for item in my_results['raw_data']]
titles = [item['title'] for item in my_results['cleaned_data']]

print(f"\n📝 Sample titles:")
for i, title in enumerate(titles[:3], 1):
    print(f"{i}. {title}")

# ========================================
# METHOD 2: Multiple Research Variables
# ========================================

# Research different topics and assign to different variables
ai_trends = await planner.research_to_variable("AI trends 2025", "standard")
tech_news = await planner.research_to_variable("Latest tech news", "simple")
market_data = await planner.research_to_variable("Tech market analysis", "deep")

# Compare results
print(f"AI Trends: {len(ai_trends['raw_data'])} items")
print(f"Tech News: {len(tech_news['raw_data'])} items") 
print(f"Market Data: {len(market_data['raw_data'])} items")

# ========================================
# METHOD 3: JSON String Assignment
# ========================================

# Get results as JSON string for easy storage/sharing
json_results = await planner.research_to_json("Quantum computing breakthroughs", "standard")

# Save to file in Colab
with open('research_results.json', 'w') as f:
    f.write(json_results)

# Load and use later
import json
loaded_results = json.loads(json_results)
print(f"Loaded {len(loaded_results['raw_data'])} items from JSON")

# ========================================
# METHOD 4: Configuration in Colab
# ========================================

# Create config dictionary for different providers
openai_config = {
    'ai_model': {
        'provider': 'openai',
        'model_name': 'gpt-4o-mini',
        'api_keys': {
            'openai_api_key': 'your-api-key-here'
        }
    },
    'research': {
        'max_sources': 10,
        'data_cleaning': {'enabled': False}  # Skip cleaning for comprehensive data
    }
}

# Apply config
for key, value in openai_config.items():
    for subkey, subvalue in value.items():
        if isinstance(subvalue, dict):
            for subsubkey, subsubvalue in subvalue.items():
                planner.config.set(f"{key}.{subkey}.{subsubkey}", subsubvalue)
        else:
            planner.config.set(f"{key}.{subkey}", subvalue)

# ========================================
# METHOD 5: Batch Research in Colab
# ========================================

async def batch_research():
    """Perform multiple research tasks in batch."""
    
    research_topics = [
        "AI in healthcare 2025",
        "Renewable energy innovations",
        "Space technology advances",
        "Quantum computing applications"
    ]
    
    results = {}
    
    for topic in research_topics:
        print(f"🔍 Researching: {topic}")
        results[topic] = await planner.research_to_variable(topic, "standard")
        
        # Show progress
        item_count = len(results[topic]['raw_data'])
        print(f"✅ Completed: {item_count} items collected")
    
    return results

# Run batch research
batch_results = await batch_research()

# Analyze batch results
for topic, data in batch_results.items():
    print(f"\n📊 {topic}:")
    print(f"   Raw items: {len(data['raw_data'])}")
    print(f"   Cleaned items: {len(data['cleaned_data'])}")
    print(f"   Success rate: {data['summary']['metadata']['successful_steps']}/{data['summary']['execution_steps']}")

# ========================================
# METHOD 6: Data Analysis in Colab
# ========================================

# Combine all collected data for analysis
all_data = []
for topic, data in batch_results.items():
    for item in data['cleaned_data']:
        item['research_topic'] = topic  # Add topic tag
        all_data.append(item)

print(f"📈 Total dataset: {len(all_data)} items across {len(batch_results)} topics")

# Create DataFrame for analysis (if pandas is available)
try:
    import pandas as pd
    
    # Convert to DataFrame
    df = pd.DataFrame(all_data)
    
    # Basic analysis
    print(f"\n📋 Data Summary:")
    print(f"   Unique domains: {df['domain'].nunique()}")
    print(f"   Average word count: {df['word_count'].mean():.0f}")
    print(f"   Top domains: {df['domain'].value_counts().head(3).to_dict()}")
    
    # Save to CSV
    df.to_csv('research_dataset.csv', index=False)
    print(f"💾 Dataset saved to research_dataset.csv")
    
except ImportError:
    print("📝 Install pandas for advanced data analysis: !pip install pandas")

# ========================================
# METHOD 7: Real-time Display in Colab
# ========================================

from IPython.display import display, HTML, clear_output
import time

async def research_with_progress():
    """Research with real-time progress display in Colab."""
    
    goal = "Latest developments in artificial intelligence"
    
    print("🚀 Starting research...")
    display(HTML(f"<h3>🔬 Research Goal: {goal}</h3>"))
    
    # Start research
    start_time = time.time()
    results = await planner.research_to_variable(goal, "standard")
    end_time = time.time()
    
    # Display results with HTML formatting
    html_output = f"""
    <div style="border: 2px solid #4CAF50; padding: 15px; border-radius: 10px; background-color: #f9f9f9;">
        <h2>🎉 Research Completed!</h2>
        <p><strong>⏱️ Time taken:</strong> {end_time - start_time:.1f} seconds</p>
        <p><strong>📊 Raw data items:</strong> {len(results['raw_data'])}</p>
        <p><strong>🧹 Cleaned data items:</strong> {len(results['cleaned_data'])}</p>
        <p><strong>✅ Success rate:</strong> {results['summary']['metadata']['successful_steps']}/{results['summary']['execution_steps']}</p>
    </div>
    """
    
    display(HTML(html_output))
    
    return results

# Run with progress display
research_results = await research_with_progress()