Skip to main content

Overview

Here's a quick summary of my experiments:

1. A meta-exercise in prompt architecting​

Prompt Engineering for Information Architects

A comprehensive online course designed to help IAs effectively integrate AI tools into their workflows for taxonomy design, content modeling, navigation systems, and research synthesis.

I used advanced prompting techniques to build Claude Skills that generate quality content and also audit for pedagogical effectiveness. The audit results are automatically synced to Notion databases, and from there to GitHub for trackable bug fixes.

This project is what I make of the future of AI-augmented content operations, turning what could be months-long development into an efficient, quality-assured system.

[Claude Skills] → [Audit Process] → [Notion Database] → [GitHub Sync]
    ↓                   ↓                  ↓                ↓
Content Generation → Quality Checks → Results Logged → Version Control

Details in my blog.

2. Card sorting prompt demo​

Prompt library for IAs

59 reusable prompts for information architecture work, extracted from the Prompt Engineering for IAs course.

WIP; I'm figuring out how to present a complete and interactive prompt library efficiently. For now, you can test the card sort prompt with an OpenAI API key, or study the simulation to understand how it works.

3. An ontology and RAG demo using unstructured content​

AI for IA Ontology and RAG

A RAG (Retrieval Augmented Generation) system that uses unstructured MDX source files with a structured JSON ontology layer.

This project is about retaining the Prompt Engineering for IAs (aka AI for IA) course content in Docusaurus-friendly format while enabling it to also be a machine-readable knowledge base.

Details in my blog.

4. An interactive web app from a PDF​

Ikigai: Find your Purpose

An interactive React-based tool that guides users through the (Japanese) concept of Ikigai to find their purpose.

I converted a PDF (and complex philosophical concepts) into intuitive user experiences using GitHub Copilot while leveraging Claude for rapid prototyping and problem-solving.

Details in my blog.