limcheekin / serverless-flutter-gpt

FlutterGPT - AI chatbot powered by OpenAI API, Qdrant, LangChain and AWS Lambda

🤖 FlutterGPT 🤖

An AI chatbot that can answer questions about Flutter. It is powered by OpenAI API and LangChain framework, and hosted in AWS Lambda and Qdrant Cloud. Feel free to chat with the bot via Telegram.

Therefore, for the correct operation of the chatbot, it is necessary to:

• get OpenAI API key > https://platform.openai.com/account/api-keys. You may be granted some free credits.
• create an AWS account > https://portal.aws.amazon.com/billing/signup, free tier plan is good enough. Also, create access key at https://us-east-1.console.aws.amazon.com/iamv2/home#/security_credentials/access-key-wizard for AWS Lambda deployment.
• create Qdrant account on > https://cloud.qdrant.io/ then create a cluster for vector store.

📁 Table of Contents

• General Info
• Technologies Used
• Features
• Requirements For Initial Setup
• Setup
• Contact

ℹ️ General Information

• The chatbot's knowledge is based on information from the Flutter documentation website, which is hosted at https://docs.flutter.dev/ and was last updated in March 2023. The static HTML files are stored in the site directory.
• Generate embeddings from the static HTML files using the OpenAI API and save them to Qdrant Cloud.
• Utilize the Serverless Framework to deploy the handler.py code to AWS Lambda.
• Access the Lambda URL to ask questions and receive responses.
• The blog post "ChatGPT Over Your Data" provides a clear explanation of how the chatbot operates behind the scene.

💻 Technologies Used

• langchain
• openai
• qdrant_client

🌟 Features

• Using OpenAI API and language model (ChatGPT) with custom knowledge base.
• Potential multi-channels support

👀 Requirements For Initial Setup

• Install Python, should work with any python version 3.9 and above
• NodeJS and Serverless Framework is not needed unless you want to deploy the code and run serverless commands from your own PC. It is optional.
  • Install NodeJS, should work with any node version above 16.16.0
  • Install Serverless Framework

📟 Setup

1. 💾 Clone/Download the Repository

2. 📦 Create Virtual Environment and Install Dependencies:

$ cd serverless-flutter-gpt
$ python -m venv venv # create virtual environment "venv"
$ source venv/bin/activate # activate it
$ pip install -r requirements.txt

3. 📔 Setup environment variables

Append the following environment variables to the venv/bin/activate file:

export QDRANT_URL = "<The URL of the cluster in Qdrant cloud>"
export QDRANT_API_KEY= "<The API key of the Qdrant cloud account>"
export OPENAI_API_KEY = "<The API key of the OpenAI API>"

Re-run the following command to activate environment variables:

$ source venv/bin/activate

Notes:

Remember to configure the environment variables above manually to Lambda once it is deploy successfully as the step is not implemented in continuous deployment.

4. 🆕 Ingest Data (Create Embeddings)

$ python create_embeddings.py

5. 🚀 Continuous Deployment to AWS Lambda

The code of querying API handler.py of the chatbot is continuous deploy and test by GitHub Action using Serverless Framework on every git push. Please look into the .github/workflows/dev.yml file to find out more.

Credentials is needed for AWS deployment, add AWS_KEY and AWS_SECRET to the repository secrets at https://github.com/<username>/<repo>/settings/secrets/actions.

💬 Contact

Created by @limcheekin - feel free to contact me!