Skip to content

Cookbook

Hands-on, runnable recipes for building with AI on Google Cloud. Each cookbook is a self-contained series that takes you from fundamentals to a deployed application — written agents-first, so both AI agents and human developers can follow along.

Available cookbooks

Gemini Cookbook

Build practical applications with Google Cloud's Gemini models — from a basic Streamlit chatbot to Retrieval-Augmented Generation and knowledge graphs, all deployable to Cloud Run.

Open the Gemini Cookbook →

A2A Cookbook (coming soon)

Recipes for the Agent-to-Agent (A2A) protocol: exposing agents over A2A, wiring multi-agent systems, and interoperating across frameworks. Coming soon — watch this space (or the blog).

What to expect

  • Practical first — every lesson ships runnable code.
  • Clear and incremental — each lesson builds on the last.
  • Production-minded — deployment, cost, and safety are treated as first-class, not afterthoughts.

Prerequisites

  • Basic Python.
  • A Google Cloud account with billing enabled and Vertex AI APIs on.
  • Comfort with the command line and git.