LouieAI Python Client

AI-powered investigation platform for natural language data analysis.

louie.__.graphistry.flow_.python.edition.mp4

Video: Louie <> Graphistry - Python edition!

🚀 Get Started in 30 Seconds

# 1. Install
# pip install louieai

# 2. Authenticate with your PyGraphistry server
import graphistry
graphistry.register(
    api=3,
    server="hub.graphistry.com",  # or your enterprise server
    username="[email protected]",   # your actual username
    password="<your-password>"    # your actual password
)

# 3. Start analyzing with natural language
from louieai.notebook import lui
lui("Show me all suspicious transactions from the last week")

# 4. Access the results
print(lui.text)  # Natural language explanation
df = lui.df      # Pandas DataFrame with the data

Install & Authenticate

pip install louieai

Authentication

LouieAI uses PyGraphistry for authentication. You'll need a free account:

1. Get PyGraphistry credentials at hub.graphistry.com (free signup)
2. Set environment variables or authenticate in code:

# Option 1: Environment variables (recommended for notebooks/scripts)
export GRAPHISTRY_USERNAME="[email protected]"
export GRAPHISTRY_PASSWORD="Analytics2024!"
export GRAPHISTRY_SERVER="hub.graphistry.com"  # or "my-company.graphistry.com"

# Optional: Custom Louie endpoint (defaults to https://louie.ai)
export LOUIE_URL="https://louie-enterprise.company.com"

# Option 2: Authenticate in code
import graphistry
graphistry.register(
    api=3, 
    server="hub.graphistry.com",  # or your enterprise server
    username="[email protected]", 
    password="password123"  # example password
)

# Optional: Use custom Louie server
from louieai import LouieClient
client = LouieClient(
    server_url="https://louie.ai",  # Louie service endpoint (default)
    server="hub.graphistry.com"      # PyGraphistry server (default)
)

Quick Start

# First, authenticate with your PyGraphistry server
import graphistry
graphistry.register(
    api=3,
    server="hub.graphistry.com",  # or your private server
    username="[email protected]",
    password="password123"  # example password
)

# Now import and use LouieAI (two options)

# Option 1: New clean import (recommended)
import louieai
lui = louieai()  # Automatically uses PyGraphistry auth

# Option 2: Traditional import
from louieai.notebook import lui

# Ask questions in natural language  
lui("Find accounts sharing payment methods or shipping addresses")

# Get fraud insights instantly
print(lui.text)
# Output: "Found 23 suspicious account clusters sharing payment/shipping details:
# 
# **Payment Card Sharing**:
# • Card ending 4789: Used by 8 different accounts in 3 days
# • Card ending 2156: 5 accounts, all created within same hour
# 
# **Address Clustering**:
# • 123 Oak St: 12 accounts using same shipping address
# • [email protected]: 7 accounts with similar email patterns
# 
# **Risk Assessment**: 67% likely promotional abuse, 23% payment fraud"

# Access the connection data
clusters_df = lui.df
if clusters_df is not None:
    print(clusters_df.head())
    #     account_id shared_payment shared_address  cluster_size  risk_score
    # 0   user_1234      card_4789    123_oak_st            12        7.2
    # 1   user_5678      card_4789    456_elm_ave            8        6.8  
    # 2   user_9012      card_2156    123_oak_st            5        8.1

Documentation

• User Guide - Complete usage examples and tutorials
• API Reference - Detailed API documentation
• Examples - Common patterns and use cases

Links

• Louie.ai Platform - Learn about LouieAI
• PyGraphistry - Required for authentication
• Support - Report issues or get help

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

For developers: Check out DEVELOP.md for technical setup and development workflow.

License

Apache 2.0 - see LICENSE