This project aims to develop a novel Redash chat add-on that empowers your team members to extract insights from multiple Redash dashboards and connected databases using natural language interaction. The chatbot leverages Large Language Models (LLMs) to enable seamless question-and-answer conversations and autonomous knowledge discovery.
Your company is seeking to significantly enhance its data analysis capabilities, with a specific focus on comprehensive YouTube data exploration. The goal is to build a Redash chat add-on that allows your team members to extract insights from your BI platforms through natural language queries. This tool will democratize data analytics, enabling team members with non-technical backgrounds to easily explore and visualize data without deep SQL knowledge.
Conversational Queries: Users can interact with Redash using natural language queries, making the process more intuitive and user-friendly. Interactive Responses: ChatGPT generates human-like responses, providing users with informative and contextual feedback on their queries. Data Visualization: The plugin allows users to visualize query results directly within the chat interface, enabling faster data exploration and analysis. The Redash ChatGPT Plugin is an exciting project that aims to integrate natural language conversation capabilities powered by ChatGPT into your Redash dashboard. Although the plugin is still a work in progress, it currently provides functionality for engaging in conversational queries with ChatGPT directly within the Redash interface.
Local development setup Set up the prerequisites Install needed packages:
sudo apt -y install docker.io docker-buildx docker-compose-v2
sudo apt -y install build-essential curl docker-compose pwgen python3-venv xvfb
Add your user to the "docker" group
sudo usermod -aG docker $USER
Install Node Version Manager
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.3/install.sh | bash
Now log out of your desktop, then back in again, for the group change to become effective and nvm to be available.
Install NodeJS version 16 Yes, it's End Of Life. But we need to use version 16 for now until we've updated some other stuff.
nvm install --lts 16
nvm alias default 16
nvm use 16
Confirm version 16 of NodeJS is active:
nvm list
Install Yarn 1.x
npm install --global yarn@1.22.19
Clone the Redash source code and install the NodeJS dependencies
git clone https://github.com/birehan/Redash-NLP-Chatbot-Analytics
cd redash
yarn
Generate your local environment variables file Generate the .env file containing (required) environment variables, suitable for development builds:
make env
Final step Inside your .env file, add your OpenAI API key, with the name indicated below: Get your free OpenAI key - https://platform.openai.com/
OPENAI_API_KEY=***************************************** Compile and build Redash uses GNU Make to run things, so if you're not sure about something it's often a good idea to take a look over the Makefile which can help.
Build the Redash front end
make build
Build local Redash Docker image
make compose_build
On my desktop (Ryzen 5600X) that took about 12 minutes to complete the first time. After that though, it's much faster at about a minute and a half each time.
It's a good idea to check that the docker images were built ok. We do that by telling docker to show us the local "docker images", which should include these three new ones. It's important the "created" time shows them to be very recent... if it's not, then they're old images left over from something else.
docker image list
Start Redash locally, using the docker images you just built
make create_database
make up
The Redash web interface should also be available at http://localhost:5001, ready to be configured.
Once you've finished confirming everything works the way you want, then shut down the containers with:
make down