A tool towards agentic human knowledge distillation.
Levi Willms & Ken Pu — Ontario Tech University (April 2026)
| 📄 Download the Full PDF Here | 🛠️ View LaTeX Source | 💻 View PPX Codebase |
Navigating the boundaries between vast document spaces—textbooks, papers, and articles—and abstract concepts represents the grand challenge of agentic knowledge distillation. To help humanity learn faster, AI agents need specialized tools to traverse these layers. A critical missing link is the exact alignment between raw information (source text) and structured text (Markdown), which is essential for agents to translate raw data into actionable concepts and abstractions.
This thesis tackles this specific bottleneck by focusing on the navigation mechanism between documents and information. I introduce the Parsed Page eXplorer, a foundational tool that facilitates agentic learning through strict alignment between source text and Markdown. Establishing this foundation with semantic provenance empowers AI agents to seamlessly bridge the gap from documents to accessible information. Ultimately, this mechanism is designed to fundamentally transform the speed and scale of human knowledge distillation.
Citation:
@thesis{willms2026ppx,
title={Parsed Page eXplorer (PPX): A Tool Towards Agentic Human Knowledge Distillation},
author={Willms, Levi and Pu, Ken},
school={Ontario Tech University},
year={2026}
}