Cocoon revolutionizes how you engage with large datasets by introducing a RAG layer that empowers a cursor-style chatbot. Built for simplicity, Cocoon allows users to connect with various LLMs and data warehouses seamlessly, making insight extraction from extensive data pipelines easier than ever. Experience streamlined data management with a live demo or try it out in your own projects.
Cocoon is a powerful solution designed to simplify the process of building chatbots that interact with large datasets and pipelines, particularly when dealing with extensive data structures, such as those containing over 1,000 tables. It introduces a Retrieval-Augmented Generation (RAG) layer that enables seamless data management through an intuitive cursor-style chatbot interface.
Key Features
- Dynamic Chatbot Functionality: Cocoon allows you to easily access and manage your data tasks through an interactive chatbot, making complex queries more accessible.
- Live Demo: Experience Cocoon in action with the live demo using RAG for Hubspot and Salesforce data: Live Demo.
- Extensive Documentation: Discover detailed insights into all available features by visiting our Comprehensive Features Page.
Getting Started
Cocoon is straightforward to set up:
- Clean your CSV files using our Online CSV Cleaner.
- Utilize Google Colab for hands-on experience with our Data Warehouse and Data Pipeline RAG features:
Installation
Cocoon is available on PyPI. To install, use the following command after creating a virtual environment:
pip install cocoon_data -U
Connect to Your Data
Cocoon supports integration with various LLMs (such as GPT-4, Claude-3, and Gemini-Ultra) and prominent data warehouses (such as Snowflake and BigQuery). An example of establishing a connection is illustrated below:
from cocoon_data import *
# Connect to OpenAI GPT-4
openai.api_key = 'your_openai_api_key'
# Connect to Snowflake
con = snowflake.connector.connect(...)
query_widget, cocoon_workflow = create_cocoon_workflow(con)
# Display the query widget
query_widget.display()
# Start the main Cocoon workflow
cocoon_workflow.start()
Once set up, you will enjoy an enhanced experience with Cocoon as showcased in the notebook interface:
š
User Interface
For those who prefer a browser-based experience, Cocoon offers a dedicated UI for chatting over the RAG feature. To use this feature, install and run:
pip install cocoon_data -U
cocoon_data
This will initiate the Cocoon interface, allowing for efficient interaction with your data.
š
Unlock the potential of your data with Cocoon, where chatbot capabilities meet powerful data processing!