Running the examples

The examples in this library are Python programs. You run them on your own computer, and they send their questions to a model running locally in Ollama. Nothing on this website runs an example, and the pages in your browser cannot reach a model on your computer.

What runs where

Reading a page here takes no setup. Running an example needs three things on your computer.

1. Python 3.10 or newer. The books use Python 3.13.
2. Ollama, installed and running. It serves a local HTTP API at http://localhost:11434 by default.
3. The model files. The Applied AI Engineering book uses qwen3:8b for chat and nomic-embed-text for embeddings. Chapter 1 explains how to download them.

Installing Python, Ollama and the model files needs an internet connection, once. After that, the local examples run without the internet.

Settings

The programs read their settings from environment variables. You do not need to change any of them to follow the book.

Variable	Default	What it sets
OLLAMA_BASE_URL	http://localhost:11434	Where Ollama is listening
OLLAMA_MODEL	qwen3:8b	The chat model
OLLAMA_EMBED_MODEL	nomic-embed-text	The embedding model

Change a variable in the terminal before you run a script, because the code reads the settings once, when it starts. The Applied AI Engineering book explains the commands for each shell.

Browser pages cannot call your model

The pages on this site are static documents. They do not run Python, and they cannot call a model on your computer. Every example is a program that you run in a terminal, and the model answers only that program.

Hosted providers are optional

Some readers will want to try a hosted model from a provider. That is optional. A chat subscription does not necessarily include API credits, API use may be billed separately, and different providers expose different features, model names and parameters. Each book explains when a hosted provider is introduced and what it changes. No core exercise requires a paid service.