What is Claude Code? Explained for non-developers
Claude Code is Anthropic's agentic coding tool, a program that writes software for you, runs it, reads the errors and fixes them until the job is done. Many people who have never written a line of code now use it for real work, so this guide explains what Claude Code is without any jargon.
One thing to clear up first is what Claude Code is not. It is not a spreadsheet macro, a file manager or a tool for merging PDFs, even if it can do all three along the way. Claude Code is a coding tool, and that matters for non-developers because software was always the gap between having an idea and having a working tool.
A team that wants someone beside them for the first weeks can see what that work looks like and costs on our Claude consultant page. Everyone else can read on and decide alone.
Claude Code in brief, checked on 9 November 2026
| Question | Short answer |
|---|---|
| What is it? | An AI agent from Anthropic that writes, runs and fixes software in your files |
| Where does it run? | Desktop app, browser, terminal, VS Code and JetBrains, Slack |
| Who is it for? | Developers first, then analysts, marketers and managers who know their field |
| Price in euros | Included from Pro at €18 a month, every Team seat includes it |
| Do I need the terminal? | No, the desktop app and the browser version need none |
What is Claude Code? The short answer
In Anthropic's own words, Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands and works with your other development tools.1 It runs in the terminal, in VS Code and JetBrains, in a desktop app for Mac and Windows and in the browser at claude.ai/code. Each of those surfaces connects to the same engine, so the settings and project notes you write once follow you between them.1
Anthropic's product page puts the idea in one line, "Hand Claude a bug fix, test, or multi-day migration".2 Underneath sits the same Claude model you may know from the chat at claude.ai, now with permission to read files, write code and run commands for you. The model family itself is explained in what Claude AI is.
Think of the difference between emailing a consultant for advice and asking the same consultant to sit down at your computer and fix the problem. The chat app is the email. Claude Code is the consultant at the keyboard, who still asks before anything risky happens.
The desktop app is the gentlest start for a non-developer. It shows every change as a visual comparison, runs several sessions side by side and can repeat a task on a schedule, and it includes Claude Code, so no separate install is needed.1
How does Claude Code work in practice?
Everything turns on the loop, the cycle of trying, checking and trying again that separates a chat assistant from an agent. A chat assistant answers one question at a time and waits for you. Claude Code takes a goal such as "fix this bug" and then works through five steps without being asked:
- Plan. Claude reads the relevant files, maps how they connect and writes a plan you can approve or change.
- Change. Claude edits the files directly, across as many of them as the task needs.
- Test. Claude runs the tests or starts the program to see whether the change works.
- Fix. When something breaks, Claude reads the error message, finds the cause and tries again.
- Report. Once the check passes, or when it gets stuck, Claude tells you what it did and what is left.
Anthropic's documentation describes the same pattern: you describe what you want in plain language, Claude plans the approach, writes the code across several files and verifies that it works.1 The difference from autocomplete tools is that Claude Code does not guess the next line and hope. It keeps working until the result matches the goal, and it shows you every file it changed before you keep anything.
Can non-developers really use Claude Code?
Yes, with one honest condition attached. Anthropic studied about 400,000 Claude Code sessions from roughly 235,000 people between October 2025 and April 2026 and published the results on 16 June 2026.3 People in software jobs reached a verified result in about 30 per cent of sessions, while people in other professions reached one in about 26 per cent.
The bigger gap was between beginners and experts in their own field. Sessions rated novice succeeded 15 per cent of the time, against 28 to 33 per cent for sessions at intermediate level or above. The authors conclude that the ability to steer Claude "comes more from command of a domain than from the ability to write code".3 A controller who knows exactly how month-end closing should work gets further than a junior developer who does not.
The authors call the findings preliminary, because a classifier judged success from transcripts and nobody measured what happened afterwards in the real world.3 Read the numbers as a direction and not a promise, which is still a striking direction for anyone who has waited months in a developer backlog.
Three changes made this possible at the same time. The models now write code that works without an expert checking every line for simple tasks, and Claude Opus 5.5 arrived on 22 September at a list price 20 per cent below Opus 5. Anthropic built the agent loop around the model, so you no longer run the tests and interpret error messages yourself. Every Team seat now includes Claude Code, Standard and Premium alike, which means a whole office can experiment and not just the developers.4
How is Claude Code different from Claude, Cowork, Cursor and Lovable?
The landscape is crowded enough that most of our first calls start with this table.
| Tool | What it is | Best for | What you get |
|---|---|---|---|
| Claude chat (claude.ai) | Chat with the Claude models | Everyone, for writing, analysis and questions | Text, documents, small apps in the chat |
| Claude Cowork | Task mode in the Claude app that works through a folder | Managers and office work | Finished documents and spreadsheets |
| Cursor | Code editor with AI built in | Developers who want suggestions as they type | Code inside an editor |
| Lovable | AI platform that builds and publishes web apps | Non-developers with an app idea | A live web app at its own address |
| Claude Code | Agent that writes, runs and fixes software | Developers and technical knowledge workers | Scripts, tools, knowledge bases, agents |
For non-developers the comparison that matters most is Lovable against Claude Code. Lovable describes itself as a platform for building, iterating on and deploying web applications, and it publishes the result to a live address straight from the project chat.5 It is quick for a prototype or a minimum viable product, meaning the smallest version that proves an idea, and it can sync the code to GitHub when you outgrow it.
Claude Code takes longer to learn and has no ceiling. You see every file that changes, connect it to whatever your company already runs and never depend on one platform's hosting decisions. Our advice to small and medium companies is simple. Try Lovable for an idea that should become a web app over a weekend, and use Claude Code for anything you will run for the next two years.
Cowork deserves its own line, because many readers of this page will be happier there. Claude Cowork works on everyday documents and folders inside the normal Claude app, so a report, a summary or a cleaned-up spreadsheet never needs Claude Code at all. Developers who compare coding agents rather than editors will find OpenAI's rival, task by task, in Claude vs Codex.
Who uses Claude Code in 2026?
Mostly developers, and there is no point pretending otherwise. Anthropic's product page lists Spotify, Stripe, Notion, Figma, Uber and NASA among the companies that use it.2 Notion's co-founder Simon Last says a big part of his job is now keeping as many Claude Code sessions busy as possible.
The interesting shift is how quickly people outside engineering pick it up. Anthropic described in July 2025 how its own teams use Claude Code, and many of the examples come from outside engineering.6 Its growth marketing team built a workflow that reads a CSV file of hundreds of ads, finds the weak ones and writes replacements in "minutes instead of hours". Data scientists who do not write TypeScript, the language behind many web apps, build whole React applications to show their results, and the legal team prototyped a system that routes colleagues to the right lawyer.
Rich Holmes at Department of Product sees the same trend from the outside, where "non-technical folks like product managers, designers and execs are dabbling with Claude Code".7 He lists SEO audits, ticket writing and personal knowledge bases among their uses. None of these people became developers. They learned to describe a result precisely and let the agent handle the syntax.
What can a non-developer build with Claude Code?
Four uses work well in European small and medium companies today, and each one starts with a task somebody already does by hand.
A writing agent that follows your rules
Our own blog runs through Claude Code. On satoriml.com we run a command called /write-blog that researches a topic, writes a draft in our house style, checks it against our SEO rules, adds the image and updates the site's index files. A post takes about 30 minutes from topic to finished draft, and a person then spends longer than that checking every fact against the vendor's own pages and approving the post before it goes live.
A knowledge base in Markdown
You have a customer wiki, a product manual or a 100-page internal handbook spread across Word files. Open Claude Code in the folder, ask it to convert everything to Markdown, a plain text format with simple headings, then add cross-links and an index. Connected to your other tools through MCP, the open standard for linking AI to data sources, the result becomes a knowledge base that both your team and Claude can read and update.1
Small scripts for the systems you already run
Your sales team wants a summary of new leads in Slack every Friday morning. Ask Claude Code to write the script, and it will ask for access to the CRM, write the code, test it against a test account and set up the weekly schedule. Somebody with admin rights still has to create the access key, and that step is where most of these projects stall for a week, so line it up first.
Understanding code nobody else understands
Your company inherited an old system and the person who built it left three years ago. Ask Claude Code to explain it, and it reads the whole codebase, maps how the parts fit together and answers your questions in plain English, often with a diagram. One afternoon of this gives you documentation that nobody had time to write in three years.
When is Claude Code the wrong tool?
Claude Code is the wrong first step when the job is a document, a summary or a slide deck, because Cowork handles those inside the normal Claude app without any project folder. It is also the wrong tool for a non-developer who wants to put customer data into production software alone. Someone who understands security still needs to review the access rights, the backups and where the data goes. A good rule from our own projects is that Claude Code may build the first version, while a developer signs off before real customers depend on it.
What does Claude Code cost in Europe?
Claude Code has no separate price, because it is included in the paid Claude plans.4 Anthropic shows European visitors these prices in euros before VAT, checked on 9 November 2026.
| Plan | Price per month | Best for |
|---|---|---|
| Free | €0 | Chat only, no Claude Code |
| Pro | €18, or €15 billed annually | One person with moderate use |
| Max | €90 | Daily professional use |
| Ultra | €180 | All-day agent work |
| Team Standard | €18 per user annually, €21.04 monthly | Companies from two seats, Claude Code included |
| Team Premium | €90 per user annually, €105.23 monthly | The heaviest users on a team |
| Enterprise | US$20 per seat plus usage, from 20 seats | SSO, audit logs and custom contracts |
Usage limits on Pro and Max are shared between the chat and Claude Code, so a heavy day in one eats into the other.8 For a five-person office on Team Standard the bill is €90 a month on annual billing, which pays for itself when each person saves one hour a month at an hourly cost above €18. Every plan and the VAT rules are in how much Claude costs, and the habits developers rely on are in our Claude Code guide with prices and setup.
Where do your files go when you use Claude Code?
Claude Code runs on your computer but sends your prompts, the files it reads and the model's answers over an encrypted connection to the model vendor.9 Under commercial terms, meaning Team, Enterprise and the API, Anthropic does not train its models on that content. Anthropic's own plans process the data outside the EU. A company that must keep processing in Europe can point Claude Code at Amazon Bedrock with EU cross-region inference instead. Called from an EU region, that route keeps each request inside six EU regions such as Frankfurt, Ireland and Stockholm.10 The wider picture for European companies is in Claude and EU data residency.
How do you get started with Claude Code?
Three routes work, depending on who you are and how many people should start at once.
One person or a small test group
Download the Claude desktop app for Mac or Windows, sign in with a Pro or Max account, open the Code tab and point it at a folder.1 Five minutes later you are working, and you never had to open a terminal. Developers who prefer the terminal install it with one line from claude.ai/install.sh and then type claude inside the project folder.
A company that wants to roll it out
A company does not want every employee installing tools on their own. Through satori-claude the licences arrive on one invoice in euros with VAT handled, for €490 once plus €14 per user a month, and the Claude licence itself costs €22 for Standard or €110 for Premium. Training runs as a Claude workshop at €280 per user for two hours from eight users.
A team that wants to test before deciding
Start with a satori-launch, where we set up Claude for your team and train everyone in a webinar and a full-day workshop. Together we find the first real tasks in your own week, for €560 per user from eight users. After that you decide with numbers in hand whether it scales.
Tip from satori.: Never start with "build me a whole app". Start with "automate this task that takes me two hours every week", because that is where non-developers get their first win fastest.
Summary and next steps
Claude Code is Anthropic's agentic coding tool, available in a desktop app, the browser, the terminal and the common code editors. It writes, runs and fixes software in a loop, and Anthropic's own data shows that knowing your field counts for more than knowing how to code. Every paid plan from Pro at €18 a month includes it, and every Team seat does too.
Pick one task that takes you two hours every week, open the Code tab in the Claude desktop app and describe the result you want in three sentences, including what a correct output looks like. Within thirty minutes you will know whether Claude Code saves you those two hours, and you will have a first script to show your team.
Sources
Footnotes
-
Anthropic (2026). Claude Code overview. Claude Code Docs, checked on 9 November 2026. https://code.claude.com/docs/en/overview ↩ ↩2 ↩3 ↩4 ↩5 ↩6
-
Anthropic (2026). Claude Code. Product page, checked on 9 November 2026. https://claude.com/product/claude-code ↩ ↩2
-
Anthropic (2026). Agentic coding and persistent returns to expertise, 16 June 2026. https://www.anthropic.com/research/claude-code-expertise ↩ ↩2 ↩3
-
Anthropic (2026). Plans and Pricing, checked from an EU connection on 9 November 2026. https://claude.com/pricing ↩ ↩2
-
Lovable (2026). Welcome to Lovable. Lovable Documentation, checked on 9 November 2026. https://docs.lovable.dev/introduction/welcome ↩
-
Anthropic (2025). How Anthropic teams use Claude Code, 24 July 2025. https://claude.com/blog/how-anthropic-teams-use-claude-code ↩
-
Holmes, R. (2025). How to use Claude Code for non-engineering use cases. Department of Product, 5 December 2025. https://departmentofproduct.substack.com/p/how-to-use-claude-code-for-non-engineering ↩
-
Anthropic (2026). Use Claude Code with your Pro or Max plan. Claude Help Center. https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan ↩
-
Anthropic (2026). Data usage. Claude Code Docs. https://code.claude.com/docs/en/data-usage ↩
-
Amazon Web Services (2026). Geographic cross-Region inference. Amazon Bedrock User Guide. https://docs.aws.amazon.com/bedrock/latest/userguide/geographic-cross-region-inference.html ↩
Frequently asked questions
Claude Code is Anthropic's agentic coding tool, meaning a program that reads your files, writes and changes code, runs commands and keeps trying until a check passes. It uses the same Claude models as the chat app and runs in a desktop app, in the browser, in the terminal and in VS Code and JetBrains.
Download the Claude desktop app for Mac or Windows, sign in with a paid plan and open the Code tab. The app includes Claude Code, so nothing else needs installing, and you can also start a session in the browser at claude.ai/code. The terminal install is one line for those who prefer it.
No, although knowing your own field matters more than you might expect. In Anthropic's study of about 400,000 Claude Code sessions, people outside software roles reached a verified result in 26 per cent of sessions against 30 per cent for software roles, while novice sessions of any background reached 15 per cent.
No, although both turn a description into working software. Lovable builds and hosts a web app from a chat in the browser, which suits a quick prototype. Claude Code works on files and folders on your own computer or in your repository, so it fits scripts, internal tools and software you will run for years.
Claude Code is included from Claude Pro at €18 a month, or €15 billed annually, before VAT. Max costs €90 and Ultra €180 a month. Every Team seat includes it, at €18 per user a month billed annually for Standard and €90 for Premium. The Free plan does not include Claude Code.
Cursor is a code editor with AI built in, so a developer stays in the editor and accepts suggestions line by line. Claude Code is an agent that takes a whole task, plans it, edits many files and runs the tests itself. Claude Code also runs inside Cursor as an extension, so a team does not have to pick one.
No, since Claude Code works on code and software projects while Cowork works on everyday documents, spreadsheets and folders inside the normal Claude app. Both run on the same models. A manager who wants reports and summaries should start in Cowork and move to Claude Code once the task needs a script or a small tool.
Claude Code runs on your computer but sends your prompts and the files it reads to the model vendor. Through Anthropic's own plans that processing happens outside the EU. A company that must keep processing in Europe can point Claude Code at Amazon Bedrock with EU cross-region inference, which keeps each request inside six EU regions when called from an EU region.
About the author
Joachim helps European companies understand and use AI in everyday work. As founder of satori., he trains teams, coaches leadership groups, and shows how tools like Claude become a natural part of work.