Claude 101 and Anthropic's Free Courses: Which to Take

Last updated: November 2026

If you are new to Claude, take Claude 101 first: it is free, it takes about two and a half hours, and it is written for people getting started with Claude at work. If you want a way of thinking about AI that will outlast whichever tool your company buys next, take AI Fluency: Framework and Foundations after it. If you are going to use Claude Code, start with Claude Code 101, because Claude Code in Action, despite sounding like the practical place to begin, is written for developers who already use the tool. And whichever you take, know in advance that two things will not be in it: what to do when Claude's instructions do not match your screen, and whether you need to learn any code first. Here is how to tell which course fits where you are.

Why trust this, and what I have not done

I teach Claude to working professionals, so I have an obvious interest in telling you that courses are not enough, and you should price that in. Against that, the verdict below is to take the free course first, before paying anyone, me included.

Here is the honest scope. This is not a review of how well these courses teach. It is a read of what each one promises, taken from its own page on Anthropic's academy in October 2026: who it says it is for, what it says you need before you start, how long it says it takes, and its lesson list. Anthropic describes the academy as "free courses, tutorials, and use cases", and each of the five courses in the table below shows "Register | FREE" on its Skilljar page. If you searched for Anthropic Academy, this is the same place under a newer name: the academy now calls itself Claude Academy, the old anthropic.com/learn address redirects to it, and the Skilljar course site still carries the Anthropic Academy label. The catalogue lists 27 courses, so I have covered only the handful a working professional is likely to need.

What I can add is the part a syllabus cannot show: what happens after the course, when someone sits down with their own machine and their own work. Two stories from teaching are below.

If you are new to Claude

Take Claude 101, and take it before anything else on this list.

Claude 101 names its audience as anyone getting started with Claude at work, and it asks almost nothing of you up front: a Claude account on any plan, including Free, and no prior experience with Claude or any other AI assistant. The page also says a few of the features it covers need a paid plan to try hands-on, which is worth knowing before you hit one and assume you are doing something wrong. If you want a free Claude AI course to start with, this is the one.

The syllabus runs 13 lessons in about two and a half hours, and the order is sensible. What Claude is, a first conversation, getting better results, then the features that turn it from a chat box into a workspace: projects, artifacts, skills, connecting your other tools, enterprise search, research for deeper questions, and a lesson on use cases by role. It ends with a quiz and a completion badge.

That list is honest about where the value sits. Most professionals I meet have used Claude as a question box, and the lessons on projects and connected tools are the jump from that to something that knows your work. They are also the part people skip when they teach themselves.

What it will not do is go deep. Thirteen topics in two and a half hours is a tour, and it is meant to be one. If your work is mostly files, folders and the same task every week, Introduction to Claude Cowork is the natural second course: written for knowledge workers who move information between files and apps, it needs a paid plan and the desktop app for the folder lessons, and no coding at all. Anthropic is now merging Cowork and chat into one Claude, starting with Pro and Max, and the course page says what you learn there still works. If you are not sure which kind of work yours is, the difference between chat, Cowork and Claude Code will tell you.

Verdict: Claude 101, because it is free, short, written for work, and assumes nothing.

If you want a framework that outlasts the tool

Take AI Fluency: Framework and Foundations, and take it second.

This is the course for the question underneath most of the questions I get: not which button to press, but how to decide what to hand an AI, how to ask for it, and how to judge what comes back. Anthropic's AI Fluency course is built around what it calls the 4D framework, Delegation, Description, Discernment and Diligence, and it was developed in partnership with two academics, Rick Dakan of Ringling College of Art and Design and Joseph Feller of University College Cork. It runs 14 lessons in about four hours, and you finish with a final assessment and a completion badge.

It suits working professionals because very little of it is about Claude specifically. The lessons move from why AI fluency matters and generative AI fundamentals to a closer look at each of the four Ds, with lessons on project planning, prompting technique and what it calls the description-discernment loop. That material still applies when your company changes vendors, which is a real possibility in any organisation still settling its AI policy. For a sturdier picture of why the models behave as they do, the companion course, AI capabilities and limitations, takes about three and a half hours and assumes no technical background.

The trade-off is that a framework is not a habit. You can finish this course able to name all four Ds and still paste a vague request into a chat box on Monday morning. The value arrives only when you apply it to your own work, and no course can choose that work for you.

Verdict: AI Fluency, after Claude 101, because it teaches judgement that survives a change of tool.

If you are going to use Claude Code

Start with Claude Code 101, not Claude Code in Action.

The names suggest a beginner course and an applied one. The pages say something more specific. Claude Code 101 is written for new developers and for experienced engineers who have not yet tried a coding agent. It takes about an hour and a half across 12 lessons and covers installation, a first prompt, an explore, plan, code and commit workflow, context management, code review, and the customisation layer: the CLAUDE.md file, subagents, skills, MCP and hooks. Before you start, it asks for basic familiarity with a code editor and the command line, and a Claude account on Pro, Max or Enterprise, or an API key.

Claude Code in Action is the next step, and its page says so directly: it is for developers who already use Claude Code for single prompts and want to move to longer, less supervised, team-wide workflows, and it expects basic familiarity with Git and the command line. The nine lessons, about an hour in all, cover steering long sessions, permission modes, hooks, routines and headless runs, GitHub Actions, and verifying work you did not watch. That is valuable material, and the wrong first hour for someone who has never opened a terminal.

Notice what both pages have in common: neither teaches the command line from zero, and both expect you to arrive with it. If you do not have it yet, budget a session for it first, or read the Claude Code tutorial for beginners on this site. As a free Claude Code course pair, these two are the obvious place to start, but the best Claude Code course for a beginner is the first one, not the one with the more practical-sounding name.

Verdict: Claude Code 101 first, Claude Code in Action once you use the tool regularly, because the second course assumes the first.

If you are wondering whether to take a Python course first

Probably not a whole one. A few hours aimed at reading code, not writing it, is what I have seen work.

The clearest case I have is a QA professional who came to me able to read code and lacking the confidence to write it. I proposed about five hours of Python before we touched Claude Code. She asked for three, because she wanted to get to the AI tools, and three is roughly what she got: strings, input, arithmetic, lists, loops, the debugger, comparisons, if statements and dictionaries. She never wrote a function or a class. We moved to Claude Code in her fourth lesson. By the end of her fifth, a small game she had built with it was live on the web. In her sixth, she went from an empty folder to a live website in about half an hour.

What made those three hours worth it was her own reason for wanting them, which she gave me in her second lesson. She said she needed "to know how to read the code" to see whether there was a logic conflict in what the AI produced. The payoff showed up in her fifth lesson, when she caught a scoring bug on her first look at the live game, a story I told in full in how to learn Claude Code. That is not a programming skill. It is the habit of checking whether the logic is right, and a Claude Code course cannot give it to you, because a course has no way of knowing whether you are checking.

She also did her Python homework herself, though I had told her Claude could solve it from a screenshot, and used AI only to look up one formatting detail. The rule I gave her is the one I give everyone: ask it for guidance, not the solution.

None of this means everyone needs Python. If you are going to stay in chat and Cowork, you need none. If you are going to read what Claude Code writes, a few hours pointed at reading is enough to start, and how much Python you need in the age of AI breaks that down by goal.

Verdict: no full Python course; a few hours of basics aimed at reading code, and only if you will use Claude Code.

If you have finished a course and your screen does not match it

Then you have met the gap every course has, and the fix is a habit, not another course.

A course is recorded on someone else's machine. Your operating system, your version of the app and your company's settings are all slightly different, and instructions that were exactly right on the presenter's screen are approximately right on yours. Approximately right is where people get stuck.

The same student showed me this more clearly than anyone. Getting Claude Code to run on her Windows machine took about 51 minutes of her first lesson with it, most of it spent sending screenshots of the problem to an ordinary Claude chat. Three times across that lesson and the next, Claude gave her steps for a screen it could not see, and three times the steps were wrong. An uninstall command used an option the tool on her machine did not accept. A settings file did not contain the entry Claude told her to look for. And on the hosting platform's settings page, Claude told her to choose an option from a list where her screen had a plain text box. Each time, one screenshot fixed it. Claude looked at what was actually there and corrected itself.

Her first reaction was that it was still not that intelligent. The third time it happened, I put it differently: Claude sometimes gives you the best possible suggestion rather than the most accurate one, because without a picture it is describing a typical screen, not yours. She got there herself a moment later, saying it probably did not know the details. Half an hour into that first lesson she was already asking whether she should send the next screenshot before I could suggest it.

That loop is the most useful habit I teach in a first month, and it is not in any of the lesson lists above, because it is not a feature. It is a reflex: the moment an instruction does not match what you see, stop, capture the screen and send it, rather than guessing which button the instruction probably meant.

Verdict: no second course; build the screenshot reflex on your own machine, because no recording can see your screen.

The courses side by side

All five are free and each ends in a quiz and a completion badge. The lengths are Anthropic's own estimates, checked against the course pages in October 2026. Catalogues change faster than blog posts, so check the page before you plan a week around one.

What I tell my students

Take the free course first. It costs nothing, it is written by the company that makes the tool, and it gives you a map of features you would otherwise discover by accident over months. I say this to people who are about to pay for lessons with my team, and I mean it: arriving with Claude 101 done means the first hour goes on your work rather than a tour.

Then pick one real task from your own week and do it in the tool before you start a second course. This is where most self-directed learning stalls, and it is what the honest comparison of courses and 1-on-1 teaching is really about. A course can show you that projects exist. It cannot sit next to you while you set one up for the report you write every Friday.

That gap is where my team and I work. If Claude Code is in your future at work, learning Claude Code 1-on-1 starts from the machine and the permissions you actually have, not a demo project. And the case where I recommend against both a second course and a tutor is real: if you use Claude a few times a week for drafting and summarising and it is working, Claude 101 is all you need, and you can skip everything after it.

Frequently Asked Questions

Is Claude 101 free?

Yes. Anthropic describes its academy as free courses, tutorials and use cases, and Claude 101's Skilljar page shows "Register | FREE". You need a Claude account to take it, and the Free plan is enough, although a few of the features it covers need a paid plan to try hands-on. Signing in on the academy saves your progress. Anthropic estimates about two and a half hours; if you pause to try each feature on your own work, which is the point, budget an afternoon.

What is the difference between Claude 101 and Claude Code 101?

Claude 101 teaches the Claude app for everyday work; Claude Code 101 teaches Claude Code, Anthropic's agentic coding tool. Claude 101 is for anyone getting started with Claude at work and runs on any plan, including Free. Claude Code 101 is for new developers and engineers new to coding agents, expects a code editor and the command line, and needs Pro, Max, Enterprise or an API key. If you will never open a terminal, Claude 101 is the only one of the two you need.

Do Anthropic's free courses come with a certificate?

Each of the five courses above lists a completion badge on Anthropic's academy, and the Skilljar versions of Claude 101 and AI Fluency describe a certificate of completion. They record that you finished and passed the quiz. They are not a certification: Anthropic's certifications are separate paid, proctored exams, open only to people at Claude Partner Network organizations, and Claude 101 is one of the free prep courses for one of them, covered in what the Claude Certified Architect exam really tests. Treat the badge as proof of interest rather than proof of skill, because what counts at work is what you can do with the tool.

Which is the best Claude Code course for a beginner?

Claude Code 101, not Claude Code in Action. The second course's own page says it is for developers who already use Claude Code for single prompts, and it expects Git and the command line. Both courses list command-line familiarity as something to have before you start, so if you have never used a terminal, plan a session for that first.

Choosing between these takes about fifteen minutes once someone has seen where each one stops, and it is one of the most common first conversations we have. If you have finished Claude 101 and want to work out what to do next with your own work, bring it to a free Discovery Call.

Written by Michael Murr for AI Tutor Code: private 1-on-1 online tutoring for professionals learning Python, AI tools, Data Science, ML, and LLM engineering. 200+ students taught, 3,000+ hours delivered.