Chatbot with Flask and Pinecone

This is a simple chatbot application that uses Flask and Pinecone. It allows users to ask questions and get answers from a knowledge base.

Referred to Pinecone's example Chatbot Agents with LangChain, more examples: Examples

Getting Started

Create Pinecone Accout

Sign up free in Pinecone

Installation

1. create and activate the environment

python3 -m venv .venv
. .venv/bin/activate

2. Install the dependencies:

pip install Flask
!pip install -qU openai pinecone-client[grpc] langchain==0.0.162 tiktoken datasets

if you got error while installing pinecone-client[grpc], use the following command instead:

pip3 install -U "pinecone-client[grpc]" install -qU openai "pinecone-client[grpc]" langchain==0.0.162 tiktoken datasets

Quick start

1. Use your own API key, as changing machines or networks can easily result in getting banned replace the following API key with your own API key

YOUR_OPENAI_API_KEY = "YOUR_OPENAI_API_KEY"
YOUR_PINECONE_API_KEY = "YOUR_PINECONE_API_KEY"
YOUR_PINECONE_ENVIRONMENT = "YOUR_PINECONE_ENVIRONMENT"

2. Start the Flask application

flask --app chatbot run

The chatbot will now be running on http://localhost:5000/.

3. Training

curl --location --request PUT 'http://127.0.0.1:5000/datasheets/dst111/train' \

4. Querying

curl --location --request POST 'http://127.0.0.1:5000/chatbots/cb111/conversation_agent' \
--header 'Content-Type: application/json' \
--data-raw '{
    "query": "What is APITable?"
}'

you could change the query parameter if you have another question