Who it's for
For readers who already use ChatGPT or Claude and want an AI to work in a browser or on their computer, and for developers who want to build one. Lessons 2, 4 and 5 need no code. Lessons 1 and 3 use a terminal and an API key.
Lessons
- 1What are browser agents? AI that clicks, types and browses for youA browser agent is an AI that clicks, types and reads pages for you. See five kinds, how each sees a page, and run a first task with browser-use.
- 2ChatGPT agent tutorial: write a browser task it can finishWrite a browser task ChatGPT agent can finish, know when it stops to ask you and check the result, using OpenAI's own system card.
- 3Computer use API: let a model use a browser through your codeLet a model use a browser through your own code: the screenshot, action and result loop, OpenAI's and Anthropic's samples to run, and how to check the result.
- 4Browser agent safety: what can go wrong and what to set upWhy a web page can take over a browser agent, what OpenAI and Anthropic measured, and the habits and controls that limit the damage.
- 5Claude Cowork tutorial: let Claude use your computer, guided by your skillsSet up Claude Cowork step by step: install the app, add a plugin and a skill, turn on computer use and hand Claude a first task you can check.
Frequently asked questions
Is this browser agents course free?
Yes. Every lesson is free to read. Running the code samples in lessons 1 and 3 uses your own API key, and the vendors bill for the model calls. Lesson 5 needs a paid Claude plan.
Do I need to know how to code?
Not for lessons 2, 4 and 5. Lesson 1 has a short Python exercise, and lesson 3 is written for developers and uses a terminal.
What is a browser agent?
A browser agent is an AI system that operates a web browser for you. It reads pages, clicks, types and moves between sites to finish a task you describe, and a good one pauses to ask before a purchase or a deletion.
Is it safe to give an AI my browser?
Not without precautions. Lesson 4 covers what Anthropic and OpenAI measured and the habits that limit the damage, starting with a separate profile or a container.
Which AI do I need?
Lesson 2 is about OpenAI's ChatGPT agent and lesson 5 about Anthropic's Claude Cowork. Lessons 1 and 3 use the OpenAI and Anthropic APIs and open-source tools, and lesson 4 applies to all of them.
What should I learn after this course?
Read AI agent guardrails for sandboxes, permission rules and hooks, and Build an AI agent with the Claude API to write the tool loop yourself.