• 

  currently still a daily task generator

• 

  knowledge base

• 

  generated events

Project Story: Personal Assistant for Life Management

About the Project

In today’s fast-paced world, staying organized and managing daily tasks can be overwhelming. Inspired by this common challenge, I set out to create a personal assistant that goes beyond simple task tracking. This AI-powered assistant builds a personalized knowledge base for each individual, learning their behavior, working style, and personal habits to provide tailored assistance for life planning and organization.

The assistant generates daily and weekly plans by analyzing a user’s monthly goals, work logs, emails, and other input sources. The goal is to provide proactive, intelligent support to help users focus on what matters most. Over time, it adapts to the user’s preferences, improving its recommendations and optimizing their workflow.

Inspiration

The inspiration for this project came from my own need to balance multiple responsibilities across academics, research, and personal life. I often found myself struggling to manage deadlines and plan effectively. Existing tools, such as to-do lists or calendar apps, did not offer the level of personalization or integration I needed. So, I envisioned an AI assistant that could automatically learn from my activities and help me manage not only my tasks but also my habits and overall well-being.

Key Learnings

Building this assistant was a rewarding learning experience. I explored how large language models (LLMs) can be used for task automation, scheduling, and summarization. Additionally, I deepened my understanding of APIs like Google Calendar and Notion for seamless integration with daily workflows. I also learned about using retrieval-augmented generation (RAG) for enhancing the knowledge base, allowing the assistant to provide more accurate and contextualized recommendations.

One of the most valuable lessons was learning how to balance automation with human input. While the assistant can automate planning and provide suggestions, I realized that some user input—like logging tasks or tracking mood—was still necessary to ensure accuracy.

How I Built the Project

Technology Stack

• Google Calendar API: For syncing user tasks and deadlines with the assistant’s scheduling capabilities.
• Notion API: As the main knowledge base where the assistant stores and reads user information.
• RAG (Retrieval-Augmented Generation): To retrieve relevant data from the knowledge base and enhance the assistant's ability to generate personalized plans and summaries.
• React: For the front-end interface, providing users with an intuitive and interactive way to interact with the assistant.
• RESTful APIs: For communication between various components, such as the front end, back end, and external services.
• AWS Servers: To host the assistant’s services and handle scalability.
• Vector Database: For efficient storage and retrieval of high-dimensional data, improving the knowledge base's performance.
• Large Language Models (LLMs): Used to generate personalized plans and summaries.

The assistant reads a user’s monthly plan, breaks it down into weekly and daily plans, and tracks progress using work logs. By incorporating multiple data sources, it continuously learns from user habits to improve its recommendations over time.

Development Process

I started by integrating the Google Calendar API to enable the assistant to retrieve and schedule events. Then, I linked the Notion API to serve as the knowledge base, where the assistant can access past notes, deadlines, and other data. Using RAG with a vector database, I improved the assistant's ability to retrieve relevant information and generate more accurate and contextually appropriate summaries. The front-end was built using React to create a user-friendly interface, while the backend relies on AWS servers to ensure scalability and smooth performance.

Challenges

One of the main challenges I faced was automating the logging process. Currently, users need to manually input work logs, including their completed tasks and mood. Automating this entirely would require more advanced integrations, such as analyzing user interactions through email, social media, or wearable devices (like sleep trackers).

Another challenge was ensuring that the assistant could accurately interpret diverse data sources while personalizing its output. Since each user has unique habits and preferences, fine-tuning the assistant to adapt to different working styles remains an ongoing task.

Looking Forward

As the project evolves, I plan to implement more advanced integrations with email APIs, sleep tracking software, and potentially visual/audio models for monitoring user behavior. By continuing to enhance its ability to learn from users, the assistant will become more intelligent and capable of providing holistic life management.

This personal assistant aims to be a comprehensive tool for helping individuals achieve balance, productivity, and personal growth.

Built With

• calendar
• google
• notion
• python
• rag
• typescript