Writing & Insights
Thoughts on EdTech, AI, and Architecture.
Long-form pieces on why courses fail, how multi-agent systems change content production, and what software thinking does for curriculum design.
The Course Production Pipeline: Instructional Design + AI
A 5-stage pipeline for combining instructional design and AI in course production: humans design the architecture, AI produces inside it.
The Tool Was Never the Problem
AI changed the production layer, but swapping Storyline for a language model without a pipeline just makes bad courses faster. What actually determines quality.
AI Is a Bricklayer, Not an Architect
Why AI-generated courses fail and AI-assisted courses succeed: design the learning architecture by hand, then let AI produce inside your constraints.
E-Learning Completion Rates: Why They're Under 15% (And How to Fix Them)
The average e-learning completion rate is under 15%. Here is the real reason — and a proven framework for fixing it that does not require making your courses shorter.
The Multi-Agent Architecture for Education
Why a single LLM prompt wasn't enough — and how splitting curriculum generation into specialized agents cut production time while improving consistency across the e-learning team.
Why I Treat the Curriculum Like a Software Architecture
A curriculum has components, dependencies, and failure modes — just like a distributed system. How applying software thinking to instructional design changed the way our team ships at scale.
Why Nobody Finishes Your 1-Hour Video
In 2026, learners won't sit through an hour of anything. The fix isn't shorter videos — it's rethinking the unit of content entirely.
The Information Dump Problem
Every platform has content. The ones that actually change behavior have something else — a system for how people learn, not just what to show them.