6 Claude Features Most People Don’t Use (But Should)

by

📌 What this post covers
If you signed up for Claude but only use the basic chat, these features will completely change how you work with it. Six genuinely useful features that don’t get much attention.

1. Projects — Separate conversations by topic

By default, Claude doesn’t remember previous conversations when you start a new chat. Projects solves that.
When you create a project:

  • Conversation context carries over within the project
  • You can set fixed custom instructions per project
  • You can upload reference files once instead of re-uploading every time

How I use it:

  • Separate projects per work area (marketing / dev / HR)
  • One project per topic I’m studying (Python / English)
  • Set repeated tone/style instructions once and reuse

Where to find it: Left sidebar → Projects → New Project

2. Custom Instructions — Stop repeating yourself

If you find yourself typing “I’m a developer, answer in English, use Python for code examples” at the start of every chat, set up Custom Instructions instead.
Configure them per project, or globally in your settings.
Example instructions:

- Always answer in English
- Use Python 3.11 for code examples
- Be concise — skip the intro fluff and get to the point
- I'm a developer, so skip basic concept explanations

Set it once and it applies automatically inside that project.

3. File Upload — Let Claude read documents directly

Claude can read files, not just text you paste in.
Supported file types:

  • PDF, Word, plain text
  • Images (JPG, PNG, etc.)
  • Code files

How I use it:

  • Upload a PDF report → “Summarize the three key takeaways”
  • Upload a contract → “Flag any clauses I should be careful about”
  • Upload a screenshot → “Explain what this error message means”
  • Upload multiple files → “Compare these two documents”

Use the paperclip icon at the bottom of the chat box to attach files.

4. Artifacts — See and edit results live

When you ask Claude for code, HTML, or markdown, it renders the output in a separate panel as Artifacts.
Code runs live so you can see it working. HTML shows up as a finished webpage.
How I use it:

  • “Build a simple calculator in HTML” → working preview in the browser
  • “Visualize this data as a chart” → interactive chart you can inspect
  • “Turn this into a markdown doc” → rendered output you can iterate on

Looking at the result, you can say “change this button to blue” or “make the font bigger” and it edits in place.

5. Web Search — Pull in current information

Claude’s training data has a cutoff. Turn on Web Search and it’ll look up current information and use it in the answer.
Toggle it via the globe icon at the bottom of the chat box.
When it helps:

  • “What’s the S&P 500 doing today?”
  • “Summarize this week’s AI news”
  • “What changed in the latest React release?”

For general questions that don’t need fresh data, leave it off. Having it on can slow down responses slightly.

6. Memory — Let Claude remember you

Turn on Memory and Claude will hold onto important details from your conversations.
Tell it once “I’m a marketer” or “I’m vegetarian” and the next chat keeps that context.
Where to find it: Profile → Settings → Memory → Enable
You can review or delete what it remembered any time. Don’t feed it sensitive personal info.

✅ Quick recap

  1. Projects — Separate conversations and files by topic
  2. Custom Instructions — Set fixed directions and stop repeating yourself
  3. File Upload — Hand Claude PDFs, images, and code directly
  4. Artifacts — See code/HTML output live and iterate on it
  5. Web Search — Toggle on when you need current info
  6. Memory — Let it hold onto your context across chats

🎉 Wrapping up the series

Six posts later, we covered Claude from signup to power-user features. Getting started → avoiding mistakes → real prompts → coding workflows → picking a plan → hidden features.
The trick to getting good at Claude is just one thing. Use it a lot. It feels awkward at first, but the feel for it comes naturally as you keep talking to it.
I’ll keep posting useful AI content from here.

🔗 Full series

This post is part of an English mirror of a Korean dev experiments blog. Cross-posted for the global developer audience.

​




Leave a Comment