Anthropic Course: Claude 101 - What Have I Learned?

I have been using Claude in the free tier till now, like a normal user. Asking questions and code in the chat, getting a response, and going on to the next one.

Meanwhile, I have seen people using it effectively in many tutorial videos and blogs.

So I decided to purchase a basic plan and start experimenting with ideas.

But before that, I wanted to hear it from the people who built it and what is the productive way to actually use it?

This is what I have learned.

• The right way of creating a good prompt is by:
  • Set the stage : Role and context
  • Define the task : The action you want
  • Specify the rules : Style, tone, example

• If the conversation isn’t going in the right direction, you can ask a follow-up question, give feedback, redirect it, or start over.

• You can personalize Claude using its memory and styles features. Memory saves the context of your past conversations, while Styles let you define how Claude communicates with you (eg: concise, formal, or a custom style of your own)

• Understand common challenges you face when using AI, in terms of output, and how to mitigate them.

• It is important to build an iteration mindset - give more details, give specific feedback, and know when to start fresh.

• Learn how to collaborate with AI tools using the 4D framework: Delegation, Description, Discernment, and Diligence.

• Evaluation of your Claude workflow is important. It can be done by giving a few examples where you already know the intended output, writing test prompts to reproduce them, comparing Claude’s output against the examples you gave, and finally refining the prompts based on the gaps you found.

• Based on the type of work, you can decide whether to use Chat, Cowork, or Code in the Claude desktop app.

• You can work either locally on your system or in the cloud by connecting a GitHub repository.

• There are three modes, how claude can perform a task.

  • Manually Approve : Review and decide whether to approve each change
  • Accept Edits : Claude will apply changes without asking permission.
  • Plan : Claude creates a plan showing which files and changes are needed.

• Creating projects, using them effectively, and sharing them

• Artifacts — what they are, and how to copy, download and share them.

• Claude Skills and how to create your own.

• Skills vs. Projects — the key differences.

• Connectors and the types available.

• Enterprise Search explained, and how to set it up.

• Research - what it is, and when to use it for in-depth answers.

• What happens behind the scenes when you run Research.

  • Claude plans its approach
  • It runs multiple searches
  • It synthesises the findings
  • It provides citations so you can verify the sources

• Everyday use cases — getting the most out of Claude day to day.

Key Take aways

• The quality of what you get out depends on how much context you put in.
• A guided approach is necessary. Iterate, give feedback, and redirect when it drifts.
• Different components work together to get a refined, well-structured result.

Now, time to start experimenting. Excited to see where this goes.

.