About

AI failures aren't errors. They're shifts in the problem being solved.

Patrick Glatz is a systems designer and former educator whose work explores AI reliability, task drift, and human-AI decision making.

His experience spans more than two decades, beginning in behavioral analysis and vocational program management before a decade in K-12 instructional design. He applies that background and systems thinking to the study of AI behavior.

His research is formalized in OIS (Operational Integrity System), which uses controlled prompts to maintain task integrity. PUBS, a diagnostic model for classifying AI failure patterns, supplies the data that shapes that control.