Capstone Project for Computer Science and Software Engineering

LondonNavigator - AI Chatbot

Front-end built with React Native and expo CLI for android and ios development Backend built with Flask to connect to OpenAI, Langchain, Google Places, Pinecone and MongoDB endpoints.

Used Langchain Agents to handle maps finder, meeting scheduler and controlling the hallucinations with precise prompts.

Getting Started

First clone the repo using

git clone https://github.com/dhruwanga19/LondonNavigator.git

or use the Open with GitHub Desktop button, if you have it set up

Backend Installation

cd to backend

cd LondonNavigator/backend

Use python 3.12
Make your virtual environment using python -m venv .venv and activate it using source activate
Install the required packages using the following command:

pip install -r requirements.txt

Create an .env file to setup the environment variables

These are the variables you need to setup:

OPENAI_API_KEY = 'your openai api key'
PINECONE_API_KEY = 'your pinecone api key'
PINECONE_ENV = 'your pinecone environment name'
PINECONE_INDEX_NAME = 'your pinecone index name'
GOOGLE_PLACES_API_KEY = 'your google places api key'
MONGODB_URI = 'your mongodb uri'

Create a docs folder to store the documents for RAG from Pinecone Vector Store
- Run the ingest.py script to load all documents in the docs folder to the Vector Store
```
python ingest.py
```
- This script does systematic chunking of documents and splits documents based on size of content inside the document to ensure efficient retrieval
- Wait for the Ingestion Complete message in the console to successfully load all the documents for RAG (might take some time based on the document sizes)
Set up your Places API key.
Setup your MongoDB URI from MONGODB. Make sure to get the Python compatible URI.

🎉You are all setup for the backend!

Frontend Installation

cd to frontend

cd LondonNavigator/frontend

Install the expo modules with the following command
```
npx install-expo-modules@latest
```
In case of errors during the expo modules installation refer to the Expo docs

🎉You are all setup for the frontend!

Running the app

Running the application requires you to start the backend server and the frontend server in two different terminals

Terminal 1 run the following command in the backend/ directory with using

python app.py

Terminal 2 will run the following command in the frontend/ directory

npx expo start

🎉 Once the expo app is loaded on the server, click Get Started to go into the chatscreen and start chatting with the bot.