Getting Started with Claude: A Beginner's Guide to AI Assistants

What We’re Going to Cover

If you’ve heard about AI assistants but haven’t really used one, or you’ve used one a couple of times and walked away unimpressed, this post is for you. We’re going to walk through what Claude is, where it lives, and a few ideas that will help you get real value out of it.

No prior experience needed. No assumptions about whether you’re a developer or not.

What Is Claude?

Claude is an AI assistant made by a company called Anthropic. Anthropic was founded in 2021 and has a strong focus on AI safety, which is a fancy way of saying they care a lot about building AI that’s helpful, honest, and hard to misuse.

When you talk to Claude, you’re actually talking to one of several models in the Claude family. The main ones are:

Opus: the most capable, best for complex reasoning and long tasks
Sonnet: balanced between capability and speed, good for most everyday work
Haiku: fast and lightweight, great for quick back-and-forth

You don’t usually have to pick. Most of the time the app picks for you based on what you’re doing.

One thing worth knowing: Claude has a knowledge cutoff date. That means it knows a lot, but it doesn’t know what happened last week. If you need current information, you’ll want to tell it what you know or use tools that can look things up.

Where Claude Lives

Claude shows up in a few different places, and it helps to know which one fits what you’re doing.

claude.ai: the web browser version. This is where most people start and where 80% of everyday use happens.
Desktop and mobile apps: the same Claude, but as an app on your Mac, Windows machine, iPhone, or Android. Conversations sync between web, desktop, and mobile.
Claude Code: a version of Claude that runs in your terminal or code editor. This one is aimed at developers and can read files, run commands, and edit code on your machine.
API: for developers building their own apps on top of Claude.

Same underlying intelligence, different ways to reach it.

How to Actually Ask Claude for Something

The single biggest lever on the quality of what Claude gives you is how you ask.

A simple formula: Context + Goal + Constraints.

Context: who you are and what you’re working on
Goal: what you actually want
Constraints: format, length, tone, anything that must be included or avoided

Here’s the difference in practice.

Light on context:

Write a password policy.

Rich with context:

I’m a sysadmin at a small company with 40 employees. Draft a one-page password policy that a non-technical team can actually follow. Include minimum length, rotation rules, and how to store passwords safely.

Same topic, very different results. Specificity does the heavy lifting.

Using claude.ai (The 80% Case)

Most people can get huge value out of claude.ai alone, without ever touching the developer tools. Here’s what you’ll find there:

Conversations: start fresh, ask questions, keep iterating
File attachments: drop in PDFs, images, spreadsheets, or code and Claude can work with them
Projects: group related conversations and give them shared context
Custom instructions: standing background context that applies to every chat in a Project
Artifacts: long-form outputs like documents or code that open in a side panel you can iterate on

Projects are the most important concept here. If you find yourself asking Claude the same kind of question over and over, create a Project for it. Add a couple of reference documents. Write custom instructions that tell Claude who you are and how you want it to respond. Every conversation in that Project will start with that context already loaded.

This one habit will transform how useful Claude feels.

Teaching Claude About Your Work

If you use Claude enough, you’ll start repeating yourself. “I’m a teacher, so when I ask about lesson plans…” or “our team uses this specific format…” or “I prefer bullet points over paragraphs.”

Instead of typing that every time, you can teach Claude once.

On claude.ai, this is what Projects and custom instructions are for.

On Claude Code, there’s a file called CLAUDE.md that lives in your project folder. Claude reads it automatically every time you start a session.

Think of it like a one-page “how we work here” document that you’d hand to a new assistant. Write it once, and Claude reads it every morning.

The Extension Layer: Skills, Agents, Commands, and Plugins

Once you’re comfortable with Claude, there are four concepts worth knowing. These mostly apply to Claude Code, but understanding them helps even if you’re only using the web.

Skills

A skill is a reusable capability with its own instructions and files. If you find yourself doing the same kind of task over and over, you can package it as a skill. You invoke it by name (like /lesson-plan) and Claude follows the instructions that come with it.

Good for: the same task, repeated with variation.

Agents

An agent (or subagent) is a specialized version of Claude that gets delegated to for specific tasks. They can run in parallel and have their own focus. If you ask Claude to research something big, it might delegate that to a research agent while the main conversation keeps moving.

Good for: big research, code reviews, parallel lookups.

Slash Commands

A slash command is a quick shortcut. Built-in ones include things like /help and /clear. You can define your own for common actions you want two-keystroke access to.

Good for: quick shortcuts to common actions.

Plugins

A plugin is a bundle of skills, commands, and other capabilities packaged together. Think of it like installing an extension. Instead of setting up ten things one at a time, you install a plugin and get them all.

Good for: sharing capabilities across projects or teams.

Three Things to Try Tomorrow

Concrete takeaways you can act on without installing anything new.

Create a Project on claude.ai for something you work on every week. Add two or three reference documents. Write custom instructions that tell Claude who you are. Notice how the next conversation feels different.
Build one really good prompt for a recurring frustration. Maybe it’s status updates, meeting summaries, or explaining a technical concept to a non-technical audience. Save the prompt. Reuse it.
Attach real context instead of describing it. If you’re asking about a document, attach the document. If you’re asking about data, attach the spreadsheet. If you’re debugging something, attach the error log. Specificity turns generic output into useful output.

None of these require being a developer. All three get more valuable the longer you use them.

Where to Go From Here

If you’ve never used Claude before, start at claude.ai and just have a conversation. Ask it something you’re actually working on. Iterate. See what happens.

If you’re a developer and want to go deeper, look into Claude Code. It runs in your terminal or IDE and can read, edit, and execute code with your permission. It’s where the agent, skill, and CLAUDE.md ideas really come alive.

Looking for more learning paths beyond AI? Our roundup of 6 free resources to start learning IT today covers everything from basic digital literacy to web development and cybersecurity.

And if you want to keep talking about this stuff, join our Discord or come to one of our free weekly classes in Athens, TN. The best way to get good with AI tools is to see how other people are using them.