Taiyo Inoue has spent 16 years teaching college math. He now uses Codex to accelerate one of the most tedious parts of the job: keeping Canvas, the learning management system where students find assignments, due dates, course materials, and announcements, up to date.
For many instructors, Canvas maintenance means hours of repetitive date changes, file uploads, calendar adjustments, and announcement updates. At some Ivy League institutions, staff may handle that work. At most schools, it falls directly on faculty.
Over winter break in 2025, Taiyo realized Codex could use Canvas’s API—the software interface that lets other programs update Canvas—to handle much of the maintenance for him. He is not a programmer by trade, though he knows enough Python to describe what he wants. Codex helped him generate scripts that update course shells, shift materials to a new academic calendar, adapt to different meeting patterns, and create timely announcements.
In demos, Taiyo can start with a blank Canvas shell and populate a full semester of materials. He estimates that Codex saves him roughly four to five hours a week.
Those recovered hours changed his classroom. With less time going to Canvas upkeep, Taiyo flipped his courses: students watch lecture materials on their own time, then spend class at whiteboards solving problems together.
The format also gives him a clearer view of what students understand now that traditional take-home work is no longer a reliable signal. Taiyo wants students to retain what he calls “cognitive sovereignty”: responsibility for their own thinking and a sense that their thinking matters. Codex removes the administrative drag that made a more active classroom harder to run.
Taiyo also uses the Canvas workflow to show skeptical faculty where AI can help them. Many educators hear “AI” and think cheating, cognitive offloading, and institutional disruption. Automating date changes, file uploads, and announcements gives them back hours for teaching.
