Anthropic Academy

The courses on offer

At the time of writing the catalogue lists three free courses:

• AI Fluency: Framework and Foundations (about 4 hours, 14 lessons and a quiz). It teaches a four-part framework for working with AI: Delegation (what to hand over), Description (how to explain the task), Discernment (judging the output) and Diligence (using it responsibly).
• AI Capabilities and Limitations (about 3.5 hours, 13 lessons and a quiz). It builds a realistic mental model of language models: next-token prediction, what they know, working memory, how steerable they are and the limits of context.
• Building Effective Human-Agent Teams (beta, about 45 minutes, 5 lessons). It looks at how teams can split work between people and AI agents.

The Claude Code documentation also points to Academy courses on Claude Code, so the catalogue may be broader when you visit.

Who it suits

These are conceptual courses, not programming courses. They are useful for:

• Students who use AI assistants for study and want to avoid over-relying on them.
• Engineers and team leads deciding which work to delegate to AI and how to check it.
• Anyone who wants a vocabulary for talking about AI use with colleagues or managers.

If you want to call the API, write prompts in code or build tools for a model, this is not the right place to start.

How to take them

Start with Capabilities and Limitations if you are unsure why models sometimes fail in surprising ways. Take AI Fluency next, and apply the four Ds to a real task from your week: write down what you delegated, how you described it and how you checked the result. The human-agent teams course is short enough for one sitting once the other two are done.

Keep an assistant open while you go through the lessons and test the claims yourself. Trying a task you expect the model to fail at teaches more than reading about it.

Pair it with

• Anthropic's courses on GitHub for hands-on notebooks on the API, prompt engineering and tool use.
• Anthropic's prompt engineering documentation for current, model-specific technique.
• Andrej Karpathy's Deep Dive into LLMs if the limitations course leaves you curious about why models behave as they do.

Common questions

Is it really free? Yes, the listed courses are free to take.

Do I need to know how to code? No. These courses are about working with AI, not programming.

Do I get a certificate? The course listings do not mention one, so take the courses for the habits rather than a credential.

How long does it take? The three listed courses add up to a little over eight hours, so a week of evenings is enough.

Best for: anyone who uses AI at work or university and wants a clear, free grounding in using it well, before or alongside technical courses.