💡 Inspiration
Let's be honest: corporate policy and process documents are where joy goes to die. They are often long, dry, and notoriously difficult for humans to navigate. In the era of LLMs, they are also a nightmare for AI to parse. Messy formats and ambiguous workflows lead to massive hallucinations.
We wanted to build an automated auditor to solve this exact problem: making sure enterprise knowledge is perfectly structured for both humans and LLMs.
Our True Motivation: Aside from solving a massive corporate headache, this project is designed to get me a promotion and a fat raise. When we win, I'm buying an AMG! 🚗💨
⚙️ What it does
- Dual-Readability Auditing: Automatically scans internal documents to check if they are "LLM-friendly" and "Human-friendly".
- Automated LLM-Wiki Generation: Extracts complex internal company guidelines, summarizes them, and structures them into a searchable, interconnected "LLM-Wiki".
🛠️ How we built it
- Feishu/Lark APIs for dynamic document fetching.
- Python for preprocessing.
- OpenAI API for advanced parsing, prompt auditing, and markdown summarization.
Built With
- lark
- procedure
- prompt
Log in or sign up for Devpost to join the conversation.