Uploaded October 2026 | Updated November 2026, 2 hours ago
In this episode of AI Coding with Dr. Chuck, I begin moving several days of work on Tsugi's new organization hierarchy and LTI tool model into production.
The work includes a more complete data model for LTI 1.3 registrations, messages, placements, deployments, claims, and scopes, while also supporting LTI 1.1 tools through the same underlying model. I discuss some of the design decisions that came from comparing approaches in Sakai, Canvas, and D2L.
Then we move from design to production. I upgrade low-risk production instances, run database migrations, check that the migrations are idempotent, and perform smoke tests across Lessons, quizzes, discussions, grades, and LTI launches. I also create a release tag as a known-good rollback point before eventually merging the larger development branch into master.
Toward the end, I reflect on a recurring experience when doing substantial software development with AI: progress can be very fast until a design or implementation problem suddenly requires slowing down, understanding what went wrong, and working carefully back to solid ground.
Topics include:
Tsugi
LTI 1.1 and LTI 1.3
LTI registration and deployment
Organization hierarchies
Database design
Database migrations
Idempotent upgrades
Production deployment
Smoke testing
Release tags and rollback
PHP
Sakai
Canvas
D2L
AI-assisted software development
In this episode of AI Coding with Dr. Chuck, I begin moving several days of work on Tsugi's new organization hierarchy and LTI tool model into production.
The work includes a more complete data model for LTI 1.3 registrations, messages, placements, deployments, claims, and scopes, while also supporting LTI 1.1 tools through the same underlying model. I discuss some of the design decisions that came from comparing approaches in Sakai, Canvas, and D2L.
Then we move from design to production. I upgrade low-risk production instances, run database migrations, check that the migrations are idempotent, and perform smoke tests across Lessons, quizzes, discussions, grades, and LTI launches. I also create a release tag as a known-good rollback point before eventually merging the larger development branch into master.
Toward the end, I reflect on a recurring experience when doing substantial software development with AI: progress can be very fast until a design or implementation problem suddenly requires slowing down, understanding what went wrong, and working carefully back to solid ground.
Topics include:
Tsugi
LTI 1.1 and LTI 1.3
LTI registration and deployment
Organization hierarchies
Database design
Database migrations
Idempotent upgrades
Production deployment
Smoke testing
Release tags and rollback
PHP
Sakai
Canvas
D2L
AI-assisted software development










