Back to Resources
Prompt Library9 min read

Singapore-Context Prompting: PDPA, IRAS, MAS, EDG

How to localise Claude prompts for Singapore — PDPA terminology, IRAS tax references, MAS FEAT principles, EDG/PSG language. With working examples.

Haojun See
Haojun See

Founder & Director, On The Ground

Updated 1 May 2026

Why localisation lifts output quality

Claude is trained on global data, but its defaults skew American. Without localisation, your output will say "utilize" instead of "use", quote dollar amounts in USD by default, frame regulatory advice for US contexts, and write client-facing content in a register that lands wrong in Singapore professional services. Localisation isn't a polish step. It's the difference between AI output you can use and AI output you have to rewrite. This article gives you the building blocks for Singapore-context prompts that work across legal, financial, government, and commercial use cases.

Five Singapore-context blocks to add

Add as many of these as apply to your prompt. Most are one line each. 1. Jurisdiction and law. *"Apply Singapore law. Where Singapore law differs from UK or US, use Singapore. Where Singapore law is silent, note that explicitly rather than defaulting to other jurisdictions."* 2. Language register. *"Use Singapore commercial English. UK spelling. No Americanisms — no 'utilize', no 'going forward' as filler, no 'leverage' as a verb. Write 'Pte Ltd' not 'Inc' or 'LLC'."* 3. Currency and units. *"Currency: SGD by default. Quote both SGD and original currency for cross-border references. Use sqm for area, kg for weight, °C for temperature."* 4. Regulatory references. *"Where regulatory guidance applies, cite the specific Singapore source: PDPA / PDPC for personal data, MAS for financial services, IRAS for tax, ACRA for corporate compliance, IMDA for tech. Distinguish guidelines (advisory) from statutes (binding)."* 5. Cultural and business conventions. *"Recognise Singapore is multi-racial and multi-religious. Default to inclusive examples. Recognise Singlish as a register without belittling it. Consider Chinese New Year, Hari Raya, and other major holidays in time-sensitive recommendations."*

Specific regulatory framings

For PDPA-relevant work: *"This output may involve personal data under Singapore's Personal Data Protection Act 2012 (PDPA). Key obligations: consent, purpose limitation, accuracy, protection, retention, transfer limitation, data protection officer. When advising on personal data handling, reference PDPC guidance and distinguish core obligations from guidelines."* For MAS-regulated fintech work: *"This output may involve financial services regulated by the Monetary Authority of Singapore (MAS). Key frameworks: Technology Risk Management Guidelines, FEAT (Fairness, Ethics, Accountability, Transparency) principles for AI, MAS AI Risk Management Guidelines. See MAS news releases for recent guidance."* For IRAS-relevant tax work: *"This output may involve Singapore corporate or individual tax. Reference the IRAS e-Tax Guides (iras.gov.sg/quick-links/e-tax-guides) for current treatment. Note where treatment is unsettled or where IRAS may take a different view from the typical OECD position. CRITICAL: this is a draft aid only; final tax advice must come from a qualified tax adviser."* For EDG / PSG / EIS work: *"This output relates to Singapore government schemes: PSG (Productivity Solutions Grant — pre-approved digital solutions, up to 50% co-funding subject to approval), EDG (Enterprise Development Grant — substantial transformation projects, up to 50% co-funding for SMEs subject to approval), EIS (Enterprise Innovation Scheme — 400% tax deductions on qualifying AI expenditure subject to IRAS criteria). None are guaranteed; all require application and approval. Reference Enterprise Singapore and the Business Grants Portal."*

Putting it together

A complete Singapore-context system prompt for a legal team might look like: *"You are an AI assistant for a Singapore law firm. Jurisdiction: Apply Singapore law. Where Singapore law is silent or unsettled, note explicitly. Language: Singapore commercial English. UK spelling. No Americanisms. Use 'Pte Ltd' for Singapore companies. Confidentiality: This conversation may involve client-confidential information. Apply PDPA-aligned posture. Where I have not redacted identifiers, apply discretion in your output and avoid restating identifying details unnecessarily. Verification: For any case-law citation, mark uncertainty explicitly. Do not invent citations. Tone: Partner-level. Confident where the law is clear; candid where it is unsettled. Singapore commercial register. Now respond to the user's request."* This system prompt sets the foundation for every interaction. Adapt the audience and purpose for your team.

Where Singlish fits

For consumer-facing tools serving Singapore retail customers (F&B, retail, transport), recognising Singlish is useful. Add: *"Customers may write in Singlish or code-switch between English, Mandarin, Malay, or Tamil. Recognise Singlish discourse particles ('lah', 'leh', 'lor') as register markers, not errors. Respond in clear English unless the customer's preference indicates otherwise. Don't mock or correct Singlish — it's a legitimate variety."* For formal legal, financial, or B2B output: standard Singapore commercial English. Don't introduce Singlish into formal contexts.

Where to take this

For role-specific prompts that already include Singapore context, see: - Legal: AI Contract Review in Singapore, PDPA-Safe Claude Prompts - SME: Claude for Singapore SMEs: 10 Prompts - Audit: Claude for Auditors - CFO: Claude for CFOs For the full prompt library across functions, see The Singapore Prompt Library. For PDPA framework, see PDPA Prompting Checklist. OTG runs custom Singapore-context training for SG firms. Book a free 30-minute call to discuss.

Frequently asked questions

Why does context matter so much?

Generic Claude prompts produce generic American-English output that misses Singapore conventions. With localised prompts, output is sharper, more accurate to local regulation, and reads as if written by a Singapore-based team. The difference is dramatic — partner-quality vs. mediocre.

Will Claude know Singapore-specific regulations like PDPA, MAS FEAT, IRAS?

Claude has knowledge of PDPA, MAS, IRAS, and major SG regulatory frameworks at a high level. It does NOT have current case law, recent guidance updates, or sector-specific guidelines memorised. Always verify regulatory references against authoritative sources before relying on them.

What's the most useful Singapore context to add to prompts?

Three things: (a) jurisdiction ('Singapore law', 'PDPA-aligned'), (b) currency and units ('SGD', 'sqm not sqft'), (c) language register ('Singapore commercial English, no Americanisms'). These three additions lift output quality across most use cases.

Do I need different prompts for SG English vs. UK / US?

Slightly. Singapore commercial English is closer to UK English than US, with local conventions ('Pte Ltd', 'cheque' not 'check', 'storey' not 'story'). For client-facing output, the 'no Americanisms' instruction is usually enough.

What about Singlish — should Claude be aware of it?

For internal tools serving Singapore consumers, yes. Customer service chatbots dealing with Singlish queries should understand 'can or cannot', 'lah/leh/lor' as discourse particles, and code-switching between English and Mandarin/Malay/Tamil. Not for formal legal or financial output.

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.