Anthropic 37-Minute Free Tutorial: Building Production-Grade AI Agents, but the Video Isn't New
An Anthropic employee demonstrates in a 37-minute video how to build production-grade AI agents using a hosted platform. It's free and has Spanish subtitles. But the video is not newly released—the original was already on YouTube in May.
Today many people came across a tweet: Anthropic "just released" a 37-minute guide teaching you how to use AI agents to automate a company. It's free, presented by a Claude engineer.
Interestingly, the behavior of this tweet itself is very much like an automated agent: repost old content, add new copy, then publish. The original video went live on Claude's YouTube channel on May 26, titled 'Ship your first Managed Agent.' Someone in the comments pointed this out.
But the video itself is worth watching. It has 236,000 views and 5,200 likes on YouTube.
The instructor is Isabella He from Anthropic's Applied AI team. She walks you through building an incident investigation agent using Anthropic's Managed Agents platform. The whole demo uses six functions to get a working agent running.
The platform has three core concepts:
- Agent: the brain of the agent, defining the model, MCP servers, and skills.
- Environment: the hands of the agent, a secure container in which the agent performs actions.
- Session: connects the brain and the hands; one session corresponds to one running agent instance.
The agent loop runs on Anthropic's server side, so you don't have to manage hosting and scaling yourself. The design decouples the brain and the hands, making credentials more secure and reducing first-token latency—the video mentions P95 latency is reduced by more than 90%.
It has Spanish subtitles. Online comments say: '37 minutes of content, worth months of practice.' 'Worth more than half of paid courses.'
If you're working on AI agents, or want to understand what hosted agents can actually do, take a look. Just don't be misled by 'just released.' The original video is on YouTube:
发布时间: 2026-11-21 18:02