Personal AI Assistant: Architecture and Features

This project implements a modular Personal Assistant System capable of handling a variety of tasks such as email management, meeting scheduling, PDF document parsing, and web search functionality.

Getting Started

Prerequisites

• Python 3.x
• Required Libraries: PyPDF2, jproperties, requests, and others listed in requirements.txt.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/personal-assistant-system.git
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up API credentials for Gmail, Google Calendar, Brave API, and the language model.

Usage

1. Run the main script with appropriate commands:

   python main.py "Schedule a meeting with John on Monday at 2 PM"
   

---

Overview

The system consists of:

• A Main Script that serves as the entry point, routing user inputs to the appropriate services.
• Service Modules that handle domain-specific tasks like email and calendar management, PDF processing, and web search.
• A Language Model Integration to process natural language queries and classify commands.

Architecture Components

1. Core Components

• Main Script (main()):

  • Entry point for the application.
  • Parses command-line inputs to determine the appropriate service to execute.

• LLMService:

  • Connects to an external language model (e.g., llama3-8b-8192).
  • Interprets and classifies user inputs.
  • Provides utilities for identifying whether a query involves personal or public data.

---

2. Service Modules

• GSuiteService:

  • A subclass of LLMService for handling email and calendar-related tasks.
  • Integrates with Gmail API for:
    • Sending, replying to, and reading emails.
  • Utilizes Google Calendar API for:
    • Scheduling meetings, parsing attendee information, and creating events.
  • Includes logic to determine if emails require responses.

• PdfService:

  • Extracts text from PDF documents using PyPDF2.
  • Processes text with the language model for:
    • Summarizing content.
    • Answering user queries based on document information.

• SearchService:

  • Performs web searches using the Brave API.
  • Retrieves and summarizes results efficiently with multithreading.

• Groq Client:

  • Integrated within LLMService for interacting with the external language model API.
  • Generates:
    • Email subjects and body text.
    • Meeting scheduling prompts.

---

3. Utilities and Configurations

• Configuration Management:

  • Uses the jproperties library to store and retrieve settings like model URLs and API keys.

• Authentication:

  • GSuiteService uses OAuth 2.0 for authenticating with Google APIs.
  • Credentials are managed using token.json or client_secret.json.

---

Key Functionalities

Command Parsing

• The task indentification is driven by the local LLM
  • Example: Commands with "message or implying message" activate GSuiteService, while "PDF or document" invokes PdfService.

Email Management

• Automates sending and replying to emails.
• Generates context-aware email content using the language model.

Meeting Scheduling

• Extracts details such as date, time, and attendees from commands.
• Creates Google Calendar events with Google Meet links.

PDF Parsing and QA

• Extracts and analyzes text from PDFs.
• Provides document summaries and answers queries based on the content.

Web Search and Summarization

• Conducts web queries via the Brave API.
• Retrieves, processes, and summarizes search results efficiently using multithreading.

---