APIs - how apps actually call models
An API (Application Programming Interface) is how software talks to the model directly, instead of a human typing into a chat window. When an app "uses AI," it's almost always sending a request to a model's API behind the scenes and getting a response back to display or act on.
Every API call is a fresh, stateless request (see A3) - the app itself is responsible for storing conversation history and re-sending it each time to fake "memory." This is exactly the layer you already work in when building things like Kasper Trips or ProvaCV - the model call is just one component in a larger app doing auth, storage, and UI around it.
Skip the hands-on task - you already do this. Instead: the next time you're specifying a feature to a dev (or reviewing a quote), notice whether they talk about "which model/API and what it costs per call" - that's a sign they actually understand the layer, not just gluing a chatbot widget on top.
When an app "uses AI" behind the scenes, what's actually happening?