What Is Claude Code? A Plain Guide for Non-Coders

TL;DR:

• Claude Code is Anthropic's AI agent. It reads the files in a folder, changes them, runs commands and checks its own work. A chat tool gives you answers; Claude Code carries out tasks.
• It runs in the terminal (a text window), in code editors such as VS Code and JetBrains, in the Claude desktop app, on the web at claude.ai/code and in the Claude mobile app.
• It is not free. As of 11 November 2026, the cheapest plan that includes it is Claude Pro at $20 a month, or $17 a month billed yearly.
• It can ask before every action, if you set it up that way. Start in Manual or Plan mode, work on copies of your files, and keep personal data out.
• If your work is mostly documents and spreadsheets, Anthropic's Cowork or a plain chat tool may suit you better. Claude Code is built first for software work.

Who this guide is for

This guide is for business owners, managers, marketers, trainers and admin staff who keep hearing "Claude Code" and want to know what it is. You do not need to code. You only need to know what a file and a folder are.

If you already use it, our guide to Claude Code for SEO shows one full workflow. This post stays at the starting line.

What Claude Code is

Claude Code is an AI assistant that does work inside a folder on your computer, or in Anthropic's cloud. You describe a task in everyday English. It makes a plan, opens files, writes new ones, runs small programs and checks the result. It shows you what it is about to do, and you approve or refuse.

The key word is agent. A chat tool gives you an answer, and you do the work. An agent does some of the work itself, step by step.

Say you have 40 photos from a training day, named IMG_4471.jpg and so on.

• A chat tool explains how to rename them. You rename 40 files by hand.
• Claude Code looks at each photo, proposes a sensible name, asks your permission and renames them.

Where Claude Code runs

Every version uses the same engine, so your settings follow you. As of November 2026:

If you are not a coder, start with the desktop app. Anthropic's terminal guide points beginners there, because it lets you skip the terminal entirely. The app already includes Claude Code, but it needs a paid plan.

What it can do beyond code

Claude Code was built for software engineers, but it can read and write almost any file, so it also handles office chores.

One feature is worth knowing by name: CLAUDE.md. It is a plain text file in your folder that Claude Code reads at the start of every session. Put your rules there ("Use British spelling. Never delete files. Save new files in Output.") and you stop repeating yourself. Our guide on how to write SOPs for AI shows how to write rules that give the same result every time.

Where it is the wrong tool. For drafting an email or brainstorming, a chat tool needs less setup. And Claude Code knows nothing about your business beyond the folder and what you tell it.

Claude Code, Cowork or plain chat?

Anthropic now sells a separate agent called Cowork for office work. Its product page says Claude Code is built for software engineering, while Cowork is built for non-coding knowledge work such as research, analysis and document creation. Both use the same agent approach, and Cowork needs no terminal.

So for many readers, the honest answer to "should I learn Claude Code?" is "look at Cowork first."

Cowork is also paid-only: Anthropic's help centre lists Pro, Max, Ultra, Team and Enterprise (checked 11 November 2026).

Which plans include Claude Code, and the prices

These prices come from Anthropic's pricing page, checked on 11 November 2026. They are in US dollars and exclude tax. Plans change often, so check before you pay.

Two things to know before you buy:

• Every plan has usage limits. On Pro and Max, limits reset every 5 hours and every week. A long agent task uses your allowance faster than chat.
• Start with one Pro seat. Try it for a month on real work before you buy for a team.

The Kenyan reality

• Access. Kenya is on Anthropic's supported countries list, as are Uganda, Tanzania and Rwanda.
• Cost in shillings. You pay in dollars, before tax, by card. Your bank's rate and charges change the final figure, so check one month's statement before paying for a year.
• Internet. Claude Code needs a connection while it works, even on files stored on your laptop. A dropped connection stops the task.
• Your computer. Anthropic's minimum is macOS 13 or later, Windows 10 version 1809 or later, Ubuntu 20.04 or later, or Debian 10 or later, with at least 4 GB of RAM.

How it stays safe

Claude Code can change and delete files, so set up safety first.

Permission modes

A permission mode decides what Claude Code may do without asking. The main ones, from Anthropic's documentation as of November 2026:

Watch the default. Anthropic's security page says Auto is now the starting mode for new terminal and VS Code sessions on recent versions. A beginner may therefore not be asked before each action. Switch to Manual or Plan at the start, or ask whoever set it up to make Manual the default. In the desktop app, check the mode selector before your first message. Anthropic is clear that reviewing proposed commands before you approve them is your job. Read each prompt before you press Yes.

Undo, with limits

Claude Code saves a checkpoint each time you send a prompt. Type /rewind, or press Esc twice on an empty prompt, to restore files to an earlier point. But rewind only covers files changed by its own editing tools. A file changed only by a command it ran, or by you outside Claude Code, is not tracked. Anthropic says checkpoints do not replace version control, so keep a real backup.

Your data

Claude Code sends your prompts, and the content it works with, to Anthropic's servers over an encrypted connection. As of November 2026, Anthropic's data usage page says:

• On Free, Pro, Max and Ultra (consumer plans), you choose whether your data trains future models. If you allow it, data is kept for 5 years; if not, for 30 days.
• On Team, Enterprise and the API (commercial terms), Anthropic does not train on your code or prompts unless you opt in. Standard retention is 30 days.

If a folder holds client records, staff files or anything covered by Kenya's Data Protection Act, do not point Claude Code at it without a company plan and a written policy. Our guide to an AI usage policy for Kenyan organisations lists what staff should never give any AI tool.

Prompt injection

Prompt injection is when a web page, document or email hides instructions aimed at the AI, such as "send this file to this address." Anthropic advises reviewing commands before approval and not feeding untrusted content straight into Claude, and warns that no system is fully immune. Keep Manual mode on whenever it reads from the internet.

Your first hour: five safe tasks

Make a folder called "Claude Practice" and copy a few harmless files into it. Open Claude Code in that folder (in the desktop app, pick the folder when you start a session) and choose Plan or Manual mode.

1. Ask what it sees

Look at the files in this folder and tell me what each one is, in one line each. Do not change, move or delete anything.

2. Rename files

Rename the photos in this folder so each name describes the picture, in lowercase words joined by hyphens. Show me the full list of old and new names first and wait for my approval.

"Show me first and wait" is a good habit for every task that changes things.

3. Combine two spreadsheets

Change the parts in square brackets to your own file names.

There are two CSV files in this folder: [file name 1] and [file name 2]. They have the same columns. Combine them into a new file called combined.csv and remove exact duplicate rows. Tell me how many rows each file had and how many the new file has. Do not edit the original files.

The row counts let you check its work in seconds.

4. Write your house rules

Create a file called CLAUDE.md in this folder with these rules: use British spelling; never delete a file without asking me; save every new file in a folder called Output; end each task with one sentence on what you changed.

5. Build something small

Make a one-page web page called checklist.html with a checklist for preparing a training room: projector tested, chairs set, attendance sheet printed, water ready. Use large, clear text. Then tell me how to open it.

Double-click the file to open it in your browser.

Then try one task again with a vaguer prompt. The result gets worse, and that is the lesson: clear instructions, clear limits, and a check at the end.

What goes wrong, and how to avoid it

Who should not bother yet

Claude Code costs money and time to learn. Skip it for now if:

• You only write emails and reports. A free chat tool gives you most of the value. Our list of free AI tools for small businesses in Kenya is a better start.
• You cannot pay in dollars by card, or a monthly subscription does not fit your budget yet.
• Your work lives in systems you cannot export, such as a locked HR or banking platform. If it cannot reach your data, it cannot help.
• You handle sensitive data and have no AI policy. Write the policy first.
• You will not check its work. An agent that acts fast can make mistakes fast.

None of this is permanent. Many people move to an agent later, when they meet a task chat cannot do.

How it compares with Codex, GitHub Copilot and Cursor

Claude Code is one of several AI coding agents. Each has real strengths, and we do not rank them. This table uses each company's own pages, checked 11 November 2026.

Our opinions, for what they are worth:

• Codex makes most sense if your team already lives in ChatGPT. Our guide on using Codex for SEO is written for that reader.
• GitHub Copilot has the gentlest entry point thanks to its free plan, and it sits inside editors many developers already use. GitHub lists agent mode on every plan, and the coding agent that opens pull requests from Pro upwards.
• Cursor is built around its own app, so it suits people happy to switch editors.
• Claude Code runs in many places and gives fine control over permissions.

Pick the one that fits the accounts you already pay for, and learn it properly. The habits (clear tasks, checking work, protecting data) carry across all of them.

Questions people ask

Is Claude Code free?

No. As of 11 November 2026, Anthropic's setup page says the free Claude plan does not include Claude Code. The cheapest plan that does is Pro, at $20 a month or $17 a month billed yearly. Developers can also pay per use through an Anthropic Console account.

Do I need to know how to code to use Claude Code?

No. Anthropic's guide for new users says you can describe what you want in plain English. It does help to understand files and folders, to read what it proposes before approving, and to start in the desktop app.

What is the difference between Claude and Claude Code?

Claude is the AI model and the chat app at claude.ai. Claude Code is a tool built on Claude that reads and edits files and runs commands inside a folder or in the cloud.

What is the difference between Claude Code and Cowork?

Both are Anthropic agents built the same way. Anthropic says Claude Code is for software engineering, and Cowork is for non-coding work such as research, analysis and documents. If you will not build software or edit a website, try Cowork first.

Can Claude Code delete my files?

Yes, if you allow it. In Manual mode it asks before editing files or running most commands. /rewind undoes many of its changes, but not all, so keep backups and work on copies.

Does Claude Code work in Kenya?

Yes. Kenya is on Anthropic's supported countries list as of 11 November 2026. You pay in US dollars by card, and you need a steady internet connection while it works. Anthropic's Claude Academy also has free self-paced courses, including Claude Code 101.

Sources

All sources last checked 11 November 2026.

• Anthropic, Claude Code overview
• Anthropic, Claude pricing
• Anthropic, Claude Code product page
• Anthropic, Claude Code advanced setup and system requirements
• Anthropic, Terminal guide for new users
• Anthropic, Configure permissions
• Anthropic, Permission modes
• Anthropic, Claude Code security
• Anthropic, Checkpointing
• Anthropic, Data usage
• Anthropic, Use Claude Code in the cloud
• Anthropic, Cowork product page
• Anthropic Help Centre, Getting started with Cowork
• Anthropic, Supported countries and regions
• OpenAI, ChatGPT and Codex documentation and Codex pricing
• GitHub, Copilot plans
• Cursor, home page and pricing

Where FuKazee fits in

FuKazee trains teams and individuals in practical AI for the work in front of them. Since 2022 we have trained more than 3,000 people. We assume nobody in the room knows AI yet, we avoid jargon, and we say plainly that nobody becomes an expert in a few days.

We never promote one tool. Most office teams start with us in the free version of a chat tool, learning the four Cs (Clarity, Context, Constraints, Customisation), because those habits carry over to every agent, Claude Code included. When a team has work that suits an agent, such as a website, data clean-up or repeated file tasks, we help them set it up safely: a practice folder, Manual mode, a CLAUDE.md with house rules, and a written policy on what data stays out.

See how our sessions work on the page for Claude AI training in Kenya.

To ask whether Claude Code, Cowork or a simpler tool fits your team, email us at lead@fukazee.com