From Terminal to Production

Go from your first terminal command to AI agents running in production.

Most AI tutorials stop at a demo. This book keeps going. Starting from zero, you learn Python, web APIs and how large language models really work. Then you build agents that use tools, remember context, retrieve knowledge and recover from their own mistakes, and you take them all the way to production.

What you will build

• A production LLM client with retries, cost tracking, effort control and guaranteed structured output

• Tool-using agents with error handling, loop detection, cost guards and human approval

• Memory and retrieval: embeddings, a vector database and retrieval-augmented generation (RAG)

• Six complete industry projects: a support ticket router, a financial compliance agent, a clinical documentation agent, a supply chain disruption monitor, a CI/CD build failure agent and an enterprise workflow orchestrator, each with its own evaluation suite

• Shared tools through the Model Context Protocol (MCP), and agents built on an agent SDK

• The production layer: an API service, containers, cloud deployment, continuous integration, monitoring and scaling

Who it is for Complete beginners who want a career building AI systems, developers from other fields who want to add agents to their skills, and technical professionals and founders who need to understand how agents are built, where they fail and what they cost.

No mathematics beyond arithmetic and no prior programming experience are required. Every listing is complete and runnable, every chapter ends with exercises, and the code is updated for the models, prices and libraries of October 2026, with model names and prices kept in one place so you can update them in a single line.

Build things. Deploy them. Measure whether they work.