Claude Code10 min read

Claude Code in Singapore: A Non-Developer's First-Week Guide

Install, prompt patterns, PDPA tips, and your first internal tool — written for SME founders and operators who don't write code.

Haojun See

Founder & Director, On The Ground

Updated 1 May 2026

What Claude Code actually is

Claude Code is Anthropic's command-line tool that lets you describe a software task in plain English and have Claude write, edit, and run the code for you. You install it once with one command. After that, you give it goals — *"build a small dashboard that reads my Stripe data and emails me a weekly summary"* — and it goes off and does the work. It is not a chatbot. It reads your files, runs commands, edits code, runs tests, and reports back. Think of it as a junior engineer who never gets tired and works at the speed of typing. The official documentation is at docs.anthropic.com/en/docs/claude-code. Most of what you need to know is in the first ten minutes of that page.

Installing in five minutes

If you have a Mac with Node.js installed (most do), run npm install -g @anthropic-ai/claude-code in Terminal. If you don't have Node, install it from nodejs.org first. On Windows, the easiest path is to install WSL2 first. Once installed, run claude in any folder. It will ask you to log in (browser-based, no API key wrangling for the consumer plan). After that you're ready. If you'd rather have someone walk you through it in person, we run vibe coding workshops in Singapore for non-developer founders. One day, hands-on, and you'll leave with a working tool.

Your first hour: a worked example

Open Terminal in a fresh empty folder. Run claude. Then paste: *"Build a single-page web app that takes a customer inquiry (name, email, message) and uses Claude to classify it as one of: sales / support / complaint / spam. Show the classification on screen. Use Vite + React + Tailwind. Deploy-ready for Vercel."* Claude Code will: create the project structure, install dependencies, write the React components, set up Tailwind, write the classification logic against the Anthropic API, and run the dev server. You watch it work. When it's done, you have a working tool. The first time it feels like magic. The second time you realise it's not — you just stopped writing the boilerplate yourself. That's the gain. Read Vibe Coding for SMEs for the deeper why this matters.

Prompt patterns that work

Three patterns covering 80% of real use: Specify the stack. *"Use Vite + React + TailwindCSS, deploy to Vercel."* Without a stack hint, Claude will pick something reasonable but it might not match your existing infra. Be explicit. Specify the deliverable. *"At the end I should be able to run npm run dev and see it at localhost:5173."* Forces Claude to actually verify the thing works, not just write the files. Specify the constraints. *"PDPA-aligned: no data leaves Singapore-region. No analytics. No third-party JS beyond shadcn UI."* Those constraints save days of rework later. Anthropic's full prompt-engineering guide is at docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview. The XML-tag patterns described there work especially well with Claude Code.

PDPA and confidential data

Claude Code on the Pro plan does not use your inputs to train the model — that's the default for paid tiers. But the data still travels to Anthropic's servers. For most SME work that's fine. For client-confidential legal or financial work, redact identifiers before pasting, use a local LLM deployment for the sensitive parts, or talk to us about OTG Legal Box (on-device AI for law firms). For broader rules, follow our PDPA Prompting Checklist. It's eight rules; most teams break four of them.

Where to go from here

If you want to evaluate Claude Code against the alternatives, read Cursor vs Claude Code vs Copilot. If you want a list of small projects to try, read 7 Internal Tools Every SME Should Build with Claude Code. If you'd rather not learn this yourself, book a free 30-minute call — we'll scope a one-week sprint to build the tool for you, with full handover and documentation.

Frequently asked questions

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

No. Claude Code (Anthropic's command-line tool) takes plain-English instructions and writes, edits, and runs the code on your behalf. Most non-developers ship their first useful tool within a few hours of install.

How is Claude Code different from Cursor and GitHub Copilot?

Claude Code is autonomous — it executes multi-step tasks end-to-end. Cursor is a visual IDE with AI assistance. Copilot is autocomplete inside an existing editor. For 'build me an app from scratch', Claude Code wins. For day-to-day editing of existing code, Cursor or Copilot win.

Is Claude Code safe to use with confidential client data under PDPA?

Use Claude Pro / Team / Enterprise tiers, which exclude your inputs from training by default. For highly sensitive data (legal client matter, financial records), redact identifiers first or use an on-device deployment. See our [PDPA Prompting Checklist](/resources/pdpa-prompting-checklist).

How much does Claude Code cost in Singapore Dollars?

Claude Pro is approximately S$27/month per user. Claude Max is S$135–S$270/month for power users. Team plans start at S$40/user/month with a 5-user minimum. Enterprise pricing is custom.

What's the fastest first project to try?

Build an internal admin tool — a customer inquiry triage form, an invoice chaser, or a board-pack generator. We have 7 worked examples in [7 Internal Tools Every SME Should Build with Claude Code](/resources/7-internal-tools-claude-code).

PreviousWebsite as a Service: Why Singapore SMEs Are Switching to Subscriptions Next Cursor vs Claude Code vs Copilot: Which AI Code Tool Wins in 2026

Want to Apply This to Your Business?

We're a Singapore AI development and automation agency. Let's discuss how we can help solve your specific challenges.

Get in Touch View our services