From Curiosity to Assets: Building qulios.me, a Service for Organizing and Reading AI-Generated Research

Introduction

AI-powered research tools, such as OpenAI's Deep Research, are rapidly becoming mainstream. While their generative capabilities for summarizing vast amounts of information in a short time are remarkably powerful, there is still room for improvement in "how to handle reports after they are generated."

• Information Dispersion: Excellent reports end up buried deep within chat history after a few days.
• Management Noise: While many tools produce well-structured output, there is a lack of space to aggregate insights across multiple tools and accumulate/manage them as your own "intellectual assets."
• Lack of Reading Comfort: It is difficult to read reports spanning tens of thousands of characters in one sitting, and there is inadequate management of reading progress or ease of access to references.

To address this, I developed qulios.me, a completely private knowledge archive specialized in "efficient management and deep reading of AI-generated documents."

A New Way to Use "Knowledge Archiving" Provided by qulios.me

qulios.me is a place to elevate high-quality reports output by AI tools from disposable logs into "your own insights."

1. Aggregate Scattered Insights into "Your Own Specialized Library"

Consolidate research results generated from different services like ChatGPT, Gemini, and Deep Research into one place. By continuously saving them, you can build a "personal archive" that visualizes the trajectory of your interests and explorations without relying on external tools.

2. Reading Support to "Make Long Reports Part of Yourself"

With automatic synchronization of viewing position (reading percentage display), you can smoothly resume deep reading even after several days. I have created an environment where you can explore a sea of vast text without getting lost, thanks to hierarchical heading displays and automatic linking of reference markers.

3. Seamless Integration from Word/Markdown

Simply by importing Word files (.docx) or Markdown (.md) output by AI tools, they are instantly converted into an optimized viewer format.

4. Completely Private Design to Make Knowledge "Your Own Strength"

Accumulated documents are not intended for public sharing like social media, but are managed strictly as "the foundation of your own thinking." It functions as a "fortress of knowledge" where you can deepen your interests without worrying about the eyes of others.

5. Three Management Experiences to Keep Your Exploration Moving

Equipped with features to quickly access necessary information from a vast archive and resume thinking:

• My Archive: Your own encyclopedia, organizing your exploration process by timeline and category.
• Reading List: Progress management that encourages deep reading and guides incomplete explorations toward completion.
• Saved: Bookmark important past insights to return to the origin of your thoughts at any time.

---

Technical Stack Selection: The Choice of Nuxt 4 + Appwrite

In this development process, I was able to proceed smoothly from conceptualization to the launch of the MVP. In today's development environment, where AI-driven code generation has become commonplace, the decision of "how to choose an environment that does not hinder AI speed while making it easy to maintain the accuracy of generated code" was a major factor in shaping the product in such a short period.

Background of Selection

By delegating routine implementations such as infrastructure operations and authentication foundations to managed services, I was able to concentrate resources on unique business logic and the refinement of the "reading experience," which is the most important aspect.

1. Backend Integration with Appwrite

Through research, I confirmed that Appwrite is a highly rational choice for individual development.

• Integrated Ecosystem: Authentication (Magic Link) and databases are provided consistently through an SDK. Minimizing the effort of combining individual external services was one of the reasons that kept development from stalling.
• Flexibility of Data Migration: For this rapid launch, I used the cloud version; however, there is peace of mind knowing that I can easily migrate to a self-hosted environment using Docker if I want to fully own the data physically in the future.
• Intuitive Management UI: Since authentication settings and data modeling are completed through a sophisticated UI, the tempo of development was maintained, allowing me to shorten the time spent on document loading and immediately turn thoughts into form.

2. Maximizing Development Efficiency with Nuxt 4

• Auto-imports and Directory Structure: By minimizing explicit definitions of components and functions, I was able to focus on the implementation of logic.
• MDC (Markdown Components): This is a mechanism that allows embedding Vue components within Markdown. It played a major role in reflecting structured data generated by AI into rich UI elements (such as heading headers and progress gauges) as-is.

---

Allocating Resources to Implementation Quality

By delegating routine backend implementation to the service, I was able to devote resources to the details of the UI/UX, specifically "how comfortably and deeply one can read research documents."

I believe that the factor that enabled smooth progress was the selection of a flexible and agile architecture that allows AI output to be directly connected to my "personal assets."

---

Summary

How can we take conversations with AI, which tend to be "output and done," and turn them into your own assets that you can comfortably re-read?
By adopting a highly agile configuration of Nuxt 4 + Appwrite, I was able to focus on the core of this challenge: the pursuit of the "reading experience."

I hope to continue exploring new ways of engaging with "my own knowledge" in the post-AI era.

If you are interested, please try out qulios.me at least once.