Getting Started with LouieAI Notebook Interface¶

This notebook demonstrates the basic usage of the LouieAI notebook-friendly API (lui).

Authentication Requirements¶

LouieAI uses PyGraphistry for authentication. Before running this notebook, you need:

1. PyGraphistry Account: Sign up for free at hub.graphistry.com

2. Set Credentials: Use one of these methods:

   # Option 1: Environment variables (recommended for notebooks)
   export GRAPHISTRY_USERNAME="your_username"
   export GRAPHISTRY_PASSWORD="your_password"
   

   # Option 2: Authenticate in code
   import graphistry
   graphistry.register(api=3, server="hub.graphistry.com", 
                      username="your_username", password="your_password")
   

Setup¶

First, make sure you have LouieAI installed and your credentials configured.

In [1]:

import os

import graphistry

from louieai import louie

# Authenticate with credentials from environment variables
# Set these before running:
# export GRAPHISTRY_USERNAME="your_username"
# export GRAPHISTRY_PASSWORD="your_password"
# export GRAPHISTRY_SERVER="your_server"
# export LOUIE_SERVER="your_louie_server"

g = graphistry.register(
    api=3,
    server=os.environ.get("GRAPHISTRY_SERVER", "hub.graphistry.com"),
    username=os.environ.get("GRAPHISTRY_USERNAME"),
    password=os.environ.get("GRAPHISTRY_PASSWORD"),
)
print("✅ Authenticated with Graphistry")

# Create the Louie interface
LOUIE_SERVER = os.environ.get("LOUIE_SERVER", "https://den.louie.ai")
lui = louie(g, server_url=LOUIE_SERVER)
print(f"✅ Connected to: {LOUIE_SERVER}")

# Display status
lui

import os import graphistry from louieai import louie # Authenticate with credentials from environment variables # Set these before running: # export GRAPHISTRY_USERNAME="your_username" # export GRAPHISTRY_PASSWORD="your_password" # export GRAPHISTRY_SERVER="your_server" # export LOUIE_SERVER="your_louie_server" g = graphistry.register( api=3, server=os.environ.get("GRAPHISTRY_SERVER", "hub.graphistry.com"), username=os.environ.get("GRAPHISTRY_USERNAME"), password=os.environ.get("GRAPHISTRY_PASSWORD"), ) print("✅ Authenticated with Graphistry") # Create the Louie interface LOUIE_SERVER = os.environ.get("LOUIE_SERVER", "https://den.louie.ai") lui = louie(g, server_url=LOUIE_SERVER) print(f"✅ Connected to: {LOUIE_SERVER}") # Display status lui

✅ Authenticated with Graphistry
✅ Connected to: http://localhost:8000

Out[1]:

🤖 LouieAI Session

---

⚪ Session: Not started (use lui('your query'))

📚 History: 0 responses

🔍 Traces: Disabled (use lui.traces = True to enable)

Quick Help (click to expand)

# Make a query

lui('Show me sales data from last week')


# Control visibility

lui('query', share_mode='Private')       # Default: only you

lui('query', share_mode='Organization')  # Share within org

lui('query', share_mode='Public')        # Share publicly


# Access results

df = lui.df          # Latest dataframe

text = lui.text      # Latest text response

all_dfs = lui.dfs    # All dataframes


# History

lui[-1].df           # Previous response's dataframe


# Traces (AI reasoning)

lui.traces = True    # Enable for session

lui('query', traces=True)  # Enable for one query

Basic Queries¶

The lui interface accepts natural language queries and returns AI-generated responses. Let's verify it works with a simple arithmetic question:

In [2]:

# Ask a simple arithmetic question
lui("2 + 2 = ?")
print(f"Answer: {lui.text}")

# Ask a simple arithmetic question lui("2 + 2 = ?") print(f"Answer: {lui.text}")

🤖 LouieAI Response

Thread: D_NX8yJBHIrMpAHHSY8pTw | Time: 33.7s

The question is a simple arithmetic problem asking for the sum of 2 and 2.

Answer: The question is a simple arithmetic problem asking for the sum of 2 and 2.

In [3]:

# Validation: check the response
if "4" in str(lui.text):
    print("✅ Basic query working correctly!")

# Validation: check the response if "4" in str(lui.text): print("✅ Basic query working correctly!")

Working with Data¶

LouieAI can generate and analyze data. Let's create a sample dataset:

In [4]:

# Create a sample DataFrame
import pandas as pd
from IPython.display import display

sample_df = pd.DataFrame(
    {
        "name": ["Alice", "Bob", "Charlie"],
        "age": [25, 30, 35],
        "city": ["New York", "Los Angeles", "Chicago"],
    }
)

# Upload to LouieAI for analysis
lui("Describe this data", sample_df)

# Display the response
print(f"Text response: {lui.text}")
print(f"DataFrame ID: {lui.df_id}")

# Show the DataFrame using native display
if lui.df is not None:
    print("\nDataFrame returned:")
    display(lui.df)  # Native Jupyter display

# Create a sample DataFrame import pandas as pd from IPython.display import display sample_df = pd.DataFrame( { "name": ["Alice", "Bob", "Charlie"], "age": [25, 30, 35], "city": ["New York", "Los Angeles", "Chicago"], } ) # Upload to LouieAI for analysis lui("Describe this data", sample_df) # Display the response print(f"Text response: {lui.text}") print(f"DataFrame ID: {lui.df_id}") # Show the DataFrame using native display if lui.df is not None: print("\nDataFrame returned:") display(lui.df) # Native Jupyter display

Text response: DataFrame with 3 rows and 3 columns. Columns: name, age, city
DataFrame ID: B_KT2IcA3d

DataFrame returned:

	name	age	city
0	Alice	25	New York
1	Bob	30	Los Angeles
2	Charlie	35	Chicago

Analyzing Data¶

Once you have data, you can ask LouieAI to perform calculations and analysis using its code agent:

In [5]:

# Analyze the DataFrame - calculate average age using LouieAI
# Note: The server should calculate the average (30) when using the 'average' keyword
if lui.df is not None:
    # Ask LouieAI to calculate the average age - use 'average' keyword
    lui("Calculate the average of the age column", lui.df)

    print("📊 Age Analysis:")
    print(f"LouieAI response: {lui.text}")
else:
    print("No DataFrame available for analysis")

# Analyze the DataFrame - calculate average age using LouieAI # Note: The server should calculate the average (30) when using the 'average' keyword if lui.df is not None: # Ask LouieAI to calculate the average age - use 'average' keyword lui("Calculate the average of the age column", lui.df) print("📊 Age Analysis:") print(f"LouieAI response: {lui.text}") else: print("No DataFrame available for analysis")

📊 Age Analysis:
LouieAI response: The average age is 30.0

In [6]:

# Validation: Check if the age calculation was performed
if lui.text and "30" in str(lui.text):
    print("✅ Average age calculation completed!")
else:
    print("Age analysis in progress...")

# Validation: Check if the age calculation was performed if lui.text and "30" in str(lui.text): print("✅ Average age calculation completed!") else: print("Age analysis in progress...")

✅ Average age calculation completed!

Beyond DataFrames¶

LouieAI can also analyze images, PDFs, Excel files, and more!

For comprehensive examples of uploading different file types, see the Data Upload Guide:

• Image analysis (PNG, JPEG, etc.)
• Document processing (PDF, Word, PowerPoint)
• Excel file handling
• Advanced upload options

In [7]:

# Create a products DataFrame for inventory analysis
import pandas as pd

products_df = pd.DataFrame(
    {
        "product": ["Widget A", "Widget B", "Widget C"],
        "price": [10.99, 15.49, 8.99],
        "quantity": [100, 75, 150],
    }
)

print("Product inventory:")
print(products_df)

# Upload and analyze with LouieAI
# Server should calculate total as price * quantity
lui(
    "Calculate the total inventory value (sum of price * quantity for each product)",
    products_df,
)

# Show the response
print(f"\n📤 LouieAI Response: {lui.text}")

# Show the DataFrame that was uploaded
if lui.df is not None:
    print(f"\nDataFrame ID: {lui.df_id}")

# Create a products DataFrame for inventory analysis import pandas as pd products_df = pd.DataFrame( { "product": ["Widget A", "Widget B", "Widget C"], "price": [10.99, 15.49, 8.99], "quantity": [100, 75, 150], } ) print("Product inventory:") print(products_df) # Upload and analyze with LouieAI # Server should calculate total as price * quantity lui( "Calculate the total inventory value (sum of price * quantity for each product)", products_df, ) # Show the response print(f"\n📤 LouieAI Response: {lui.text}") # Show the DataFrame that was uploaded if lui.df is not None: print(f"\nDataFrame ID: {lui.df_id}")

Product inventory:
    product  price  quantity
0  Widget A  10.99       100
1  Widget B  15.49        75
2  Widget C   8.99       150

📤 LouieAI Response: The total calculated value is $3609.25

DataFrame ID: B_Ch4dXSk8

In [8]:

# Validation: Check if LouieAI returned the inventory calculation
if lui.df is not None:
    print("✅ Inventory calculation received from LouieAI!")
else:
    print("Waiting for LouieAI response...")

# Validation: Check if LouieAI returned the inventory calculation if lui.df is not None: print("✅ Inventory calculation received from LouieAI!") else: print("Waiting for LouieAI response...")

✅ Inventory calculation received from LouieAI!

Uploading Your Own Data¶

LouieAI can analyze DataFrames you create or load from files:

In [9]:

import numpy as np
import pandas as pd

# Create your own DataFrame (or load from CSV/Excel)
my_data = pd.DataFrame(
    {
        "date": pd.date_range("2024-01-01", periods=30, freq="D"),
        "sales": np.random.randn(30).cumsum() + 100,
        "visitors": np.random.randint(50, 200, 30),
        "region": np.random.choice(["North", "South", "East", "West"], 30),
    }
)

print("Sample of your data:")
my_data.head()

import numpy as np import pandas as pd # Create your own DataFrame (or load from CSV/Excel) my_data = pd.DataFrame( { "date": pd.date_range("2024-01-01", periods=30, freq="D"), "sales": np.random.randn(30).cumsum() + 100, "visitors": np.random.randint(50, 200, 30), "region": np.random.choice(["North", "South", "East", "West"], 30), } ) print("Sample of your data:") my_data.head()

Sample of your data:

Out[9]:

	date	sales	visitors	region
0	2024-01-01	99.094253	104	East
1	2024-01-02	98.690385	80	East
2	2024-01-03	101.031939	85	South
3	2024-01-04	100.059336	184	North
4	2024-01-05	101.043540	183	South

In [10]:

# Upload and analyze your DataFrame with LouieAI
lui("Calculate the total sum of the sales column", my_data)

print("📊 Sales Analysis:")
print(f"LouieAI response: {lui.text}")

# Ask a follow-up question - use 'unique' keyword for unique count
lui("Count unique values in the region column", my_data)

print("\n📊 Region Analysis:")
print(f"LouieAI response: {lui.text}")

# Upload and analyze your DataFrame with LouieAI lui("Calculate the total sum of the sales column", my_data) print("📊 Sales Analysis:") print(f"LouieAI response: {lui.text}") # Ask a follow-up question - use 'unique' keyword for unique count lui("Count unique values in the region column", my_data) print("\n📊 Region Analysis:") print(f"LouieAI response: {lui.text}")

📊 Sales Analysis:
LouieAI response: The total sum of sales is 3060.75

📊 Region Analysis:
LouieAI response: There are 4 unique values in the region column

In [11]:

# Validation: Check that LouieAI processed the regions query
if lui.text:
    print("✅ Region analysis received from LouieAI!")
else:
    print("Waiting for LouieAI response...")

# Validation: Check that LouieAI processed the regions query if lui.text: print("✅ Region analysis received from LouieAI!") else: print("Waiting for LouieAI response...")

✅ Region analysis received from LouieAI!

Session History¶

The lui interface maintains a history of your interactions:

In [12]:

# Access previous responses by index
# lui[-1]  # Last response
# lui[-2]  # Second to last response

# Check the number of interactions
print(f"Total interactions: {len(lui._history)}")

# Access previous responses by index # lui[-1] # Last response # lui[-2] # Second to last response # Check the number of interactions print(f"Total interactions: {len(lui._history)}")

Total interactions: 6

Enabling Traces¶

For complex queries, you might want to see the AI's reasoning process:

In [13]:

# Enable traces to see reasoning steps
lui.traces = True

# Now queries will show reasoning steps
lui("What is 100 + 50? Please respond with just the number.")
print(f"Answer with traces: {lui.text}")

# Disable traces
lui.traces = False

# Enable traces to see reasoning steps lui.traces = True # Now queries will show reasoning steps lui("What is 100 + 50? Please respond with just the number.") print(f"Answer with traces: {lui.text}") # Disable traces lui.traces = False

🤖 LouieAI Response

Thread: D_NX8yJBHIrMpAHHSY8pTw | Time: 5.7s

i Splunk connection successfully established.

i LouieAgent::runner

🐛 Input text:::What is 100 + 50? Please respond with just the number.

🐛 Reading chat history. (is_prepared: True)

🐛 LouieAgent predict with tools: ['GraphAgent', 'CodeAgent', 'TableAIAgent', 'FirecrawlAgent', 'NotebookAgent', 'DatabricksAgent', 'SplunkAgent', 'MermaidAgent', 'PerspectiveAgent', 'OpenSearchAgent', 'KeplerAgent', 'SpannerAgent', 'SpannerPassthroughAgent', 'KustoAgent', 'KustoPassthroughAgent', 'PostgresAgent', 'PostgresPassthroughAgent', 'Done', 'Chat'] has_recipes: True has_examples: True

150

i Louie output text: 150

Answer with traces: 150

In [14]:

# Or enable traces for just one query
lui("The sum of integers from 1 to 10 equals?", traces=True)
print(f"Sum calculation: {lui.text}")

# Validation: The sum of 1 to 10 is 55
if "55" in str(lui.text):
    print("✅ Sum formula working!")

# Or enable traces for just one query lui("The sum of integers from 1 to 10 equals?", traces=True) print(f"Sum calculation: {lui.text}") # Validation: The sum of 1 to 10 is 55 if "55" in str(lui.text): print("✅ Sum formula working!")

🤖 LouieAI Response

Thread: D_NX8yJBHIrMpAHHSY8pTw | Time: 6.6s

i Splunk connection successfully established.

i LouieAgent::runner

🐛 Input text:::The sum of integers from 1 to 10 equals?

🐛 Reading chat history. (is_prepared: True)

🐛 LouieAgent predict with tools: ['GraphAgent', 'CodeAgent', 'TableAIAgent', 'FirecrawlAgent', 'NotebookAgent', 'DatabricksAgent', 'SplunkAgent', 'MermaidAgent', 'PerspectiveAgent', 'OpenSearchAgent', 'KeplerAgent', 'SpannerAgent', 'SpannerPassthroughAgent', 'KustoAgent', 'KustoPassthroughAgent', 'PostgresAgent', 'PostgresPassthroughAgent', 'Done', 'Chat'] has_recipes: True has_examples: True

The sum of integers from 1 to 10 can be calculated using the formula for the sum of an arithmetic series: \(\text{Sum} = \frac{n}{2} \times (\text{first term} + \text{last term})\), where \(n\) is the number of terms. In this case, \(n = 10\), the first term is 1, and the last term is 10.

Calculation: \(\text{Sum} = \frac{10}{2} \times (1 + 10) = 5 \times 11 =

i Louie output text: 55

Sum calculation: The sum of integers from 1 to 10 can be calculated using the formula for the sum of an arithmetic series: \(\text{Sum} = \frac{n}{2} \times (\text{first term} + \text{last term})\), where \(n\) is the number of terms. In this case, \(n = 10\), the first term is 1, and the last term is 10.

Calculation: \(\text{Sum} = \frac{10}{2} \times (1 + 10) = 5 \times 11 =

Error Handling¶

The interface handles errors gracefully:

In [15]:

# Check if the last operation had errors
if lui.has_errors:
    print("Last operation had errors:", lui.errors)
else:
    print("No errors in last operation")

# Check if the last operation had errors if lui.has_errors: print("Last operation had errors:", lui.errors) else: print("No errors in last operation")

No errors in last operation

Tips and Tricks¶

Here are some helpful patterns for using the notebook interface:

1. Use clear prompts: Be specific about what you want to analyze
2. Check outputs: Always verify lui.df and lui.text after queries
3. Enable traces for debugging: Use traces=True when troubleshooting
4. Leverage history: Access previous results with lui[-1], lui[-2], etc.
5. Upload various formats: DataFrames, images, PDFs all work seamlessly