Strip the mystique and it is a slow, expensive, unreliable third-party call.
AI in the Stack
Where a model sits in a system, what it costs, and what it cannot be trusted with.
If you are vibe coding, you are already using models to write software. This course is about the other thing: putting a model inside the software you ship, and what that changes about your architecture.
Start the first lesson0/4 lessons read
By the end you can
- Place a model correctly in an architecture diagram
- Reason about tokens, cost, and latency before the bill arrives
- Explain why a model invents things and what actually reduces it
- Scope what an autonomous agent is permitted to touch
Lessons
The one unit of measurement that explains pricing, limits, and most weird behaviour.
Confident wrong answers are the expected behaviour, not a malfunction, and retrieval is the mitigation.
The moment a model can take actions, the interesting question stops being accuracy.