Module 12 of 12

Deploying Agents

Lessons

About This Module

An agent that works on your laptop is a prototype. An agent that real users can rely on is a service: it has to be reachable, stay up, handle many requests at once, keep its secrets safe, and run within a budget. Deploying agents is the step that turns everything you have built so far into something other people can use.

This module starts with the big picture of what it takes to move from demo to production, then looks at the common ways to host an agent: wrapping it in an API and packaging it in a container, running it on a serverless platform such as Google Cloud Run or Amazon Bedrock AgentCore, and giving it an isolated sandbox to work in, as you would for an agent built on the Claude Agent SDK.

As you watch, keep the earlier modules in mind. Hosting is only one part of production: you still need the evaluation and observability from Module 10 and the safety and guardrails from Module 11 once real traffic arrives. Some of these videos focus on a specific cloud provider, so look for the ideas that carry across platforms rather than the exact commands. When you are done, head to the projects to put the whole course into practice.

Lessons

6 videos
01

From prototype to production: Hard-won lessons for advanced, reliable, and secure agents

Looks at what it actually takes to move an AI agent from a promising demo into a high-stakes production environment, with lessons on reliability and security.

Video 42min
02

Deploying AI Agents to Production: Architecture and Infrastructure Guide

A roadmap for taking an agent from a prototype to a production-ready system, covering the architecture and infrastructure decisions along the way.

Video 8min
03

Build and Deploy AI Agents with Python, Docker, FastAPI, LangChain and LangGraph

A complete, hands-on guide to building an agent, exposing it through a FastAPI service and packaging it with Docker so it can run anywhere.

Video 4h 49min
04

Build AI agents on Cloud Run

A Google Cloud session on scaling agents beyond the prototype with serverless infrastructure, and why Cloud Run suits agent workloads.

Video 29min
05

Deploy Production-Ready Agents in 22 Minutes with AgentCore Runtime

Shows how to host framework-agnostic agents on Amazon Bedrock AgentCore Runtime, a managed platform for running agents in production.

Video 22min
06

Claude in the Box - Use Anthropic Agent SDK in a Sandbox

Shows how to host the Claude Agent SDK as a programmable agent inside a sandbox and give it access to tools, an isolated setup suited to production.

Video 8min