Module 1: How AI Systems Differ
1 min read·
What you need first
- Python (functions, classes, files), one model call made from code, and what an API, a database, a cache and a queue are.
- A Google account for Colab and Drive. Only Lesson 1 calls a hosted model (Groq or Gemini, free key); the lab and the incident file need no key.
What you will build
The start of the northdesk-platform repository: the shared data as northdesk/data.py, the lessons' probe, teaching components and call log, a failure-mode table started from the eight structural differences, and the design skeleton in the ten-step order with its decision register. It is the orientation before step 1 of the design order.
Lessons
| File | Title | Minutes | Depth |
|---|---|---|---|
01-the-structural-differences.md | Lesson 1: Eight ways an AI system breaks the old rules | 50 | Essential |
02-what-carries-over.md | Lesson 2: What still carries over from classical design | 45 | Essential |
03-the-design-posture.md | Lesson 3: Five habits that keep an AI design boring | 40 | Essential |
Extras
04-module-lab.md, Lab: The Northdesk repository and its first records (60 minutes): the repository in your Drive,%%ensure, the data, the lessons' files, two design documents and their tests.05-incident-file.md, Incident: A provider change broke a quarter of requests (30 minutes): GitHub Copilot's Gemini disruption of 29 September 2025, replayed on Northdesk's call log, with a per-model alert, a fallback and the test that checks both.