Overview

EmailAIAssistant is a desktop application developed in React and Electron. It uses gmail api integration to pull emails of a user. Once the emails are pulled it uses Llama to classify email and notifies the user if the email is important. The classification behaviour can also be trained using a chat interface.

Technologies Used

• ReactJS with Tailwind - The user interface is built using react and is styled using Tailwind CSS.

• Electron: It uses electron runtime to run on the desktop. This enables us to build and distribute the application as needed.

• Ollama - We are using Ollama to run Llama locally. This allows us to make sure that the data is not exposed to any third party.

Features

• Chat: The system is smart enough to identify what does the user want based on messages sent by the user.

Challenges and Solutions

The main challenge was to help uses stay productive while make sure that the user doesn't miss any important email. This problem also then expanded, because "important" means different things to different people. So then the challenge was to let the user also train the system in order to identify emails as per the user's preference.

Impact

EmailAssistant, illustrates that even a small model like llama 8 billion is capable in classification of emails based on the user's preference. Also it highlights our competence in developing advanced chat interfaces which serve a very specific function while making the chat seem and feel natural.