🔥 Demo: https://huggingface.co/spaces/tabularisai/augini
augini is an AI-powered data assistant that brings RAG (Retrieval-Augmented Generation) capabilities to your tabular data (CSV, Excel, XLSX). Built with state-of-the-art language models, it provides an intuitive chat interface for data analysis and powerful data manipulation capabilities.
Have natural conversations with your data using augini's chat interface. Works with any tabular format (CSV, Excel, Pandas DataFrames):
from augini import Augini
import pandas as pd
# Initialize with your preferred model
augini = Augini(api_key="your-api-key", model="gpt-4o-mini")
# Load your data (CSV, Excel, or any pandas-supported format)
df = pd.read_csv("your_data.csv") # or pd.read_excel("your_data.xlsx")
# Start chatting with your data - properly display markdown responses
from IPython.display import display, Markdown
response = augini.chat("What are the main patterns in this dataset?", df)
display(Markdown(response))
# Ask follow-up questions with context awareness
response = augini.chat("Can you analyze the correlation between age and income?", df)
display(Markdown(response))Enhance your datasets with AI-generated features:
# Add synthetic features based on existing data
result_df = augini.augment_columns(df, ['occupation', 'interests', 'personality_type'])
# Generate custom features with specific prompts
custom_prompt = """
Based on the person's age and location, suggest:
1. A likely income bracket
2. Preferred shopping categories
3. Travel preferences
Respond with a JSON object with keys 'income_bracket', 'shopping_preferences', 'travel_style'.
"""
enriched_df = augini.augment_columns(df,
['income_bracket', 'shopping_preferences', 'travel_style'],
custom_prompt=custom_prompt
)Generate privacy-safe synthetic data while preserving statistical properties:
# Define anonymization strategy
anonymize_prompt = """
Create an anonymized version that:
1. Replaces personal identifiers with synthetic data
2. Maintains statistical distributions
3. Preserves relationships between variables
Respond with a JSON object containing anonymized values.
"""
# Apply anonymization
anonymous_df = augini.augment_columns(df,
['name_anon', 'email_anon', 'address_anon'],
custom_prompt=anonymize_prompt
)pip install augini- Get your API key from OpenAI or OpenRouter
- Initialize Augini:
# Using OpenAI
augini = Augini(api_key="your-api-key", model="gpt-4o-mini", use_openrouter=False)
# Using OpenRouter
augini = Augini(api_key="your-api-key", model="meta-llama/llama-3-8b-instruct", use_openrouter=True)For enterprise deployments, local installations, or custom solutions, contact us:
- Email: info@tabularis.ai
