AI Prompt for Singapore FATF AML Gap Analysis
This prompt turns the May 2026 FATF mutual evaluation of Singapore into a firm-specific gap report. It cross-references your AML controls catalog and transaction-monitoring rules against each FATF finding, pinpoints trade-based money laundering (TBML) detection gaps, and proposes three new agentic monitoring triggers complete with threshold logic, alert-volume estimates, and the evidence trail a MAS-ready SAR requires. It is built for compliance officers and MLROs at fintech and crypto firms with a Singapore entity.
How to use this prompt
- 1
Paste the prompt into Luna in your deepidv dashboard, or into Claude, ChatGPT, or Gemini if you want a first pass without live data connections.
- 2
Attach your inputs: the Singapore entity's AML controls catalog, current transaction-monitoring rule set, investigation-to-prosecution conversion rate, open MAS or FATF advisories, and anonymized TBML-related SAR counts.
- 3
Review the GAP MAP section first: every FATF finding is marked covered, partial, or gap, with the evidence (or evidence-absence reason) stated for each.
- 4
Take the three proposed monitoring triggers to your transaction-monitoring team, pilot them at the suggested starting thresholds, and track false positives against the calibration plan.
- 5
Use the REGULATORY ENGAGEMENT section to draft your gap-closure narrative for the next MAS touchpoint, and route the flagged open questions to legal counsel.
The prompt
Luna, cross-reference our current AML controls for the Singapore entity against the May 2026 FATF review findings. Identify specific gaps in our trade-based money laundering (TBML) detection and suggest 3 new agentic monitoring triggers to improve prosecution-ready reporting. INPUT, the user will paste: - Current AML controls catalog for the Singapore entity - Existing transaction-monitoring rule set - Existing investigation-to-prosecution conversion rate - Any open MAS or FATF advisories the firm has already responded to - Recent SAR filings related to TBML (anonymized counts and outcomes) OUTPUT, return the following structured response: 1. GAP MAP For each FATF finding in the May 2026 Singapore review: - The finding (one-sentence summary, with FATF section reference) - Current firm-level coverage (covered / partial / gap) - Specific evidence of coverage or the evidence-absence reason 2. TBML DETECTION GAPS - Identify the top 3 TBML detection gaps based on the FATF assessment criteria - For each: the trade-pattern signature missed, the rule that would have fired had it existed, the volume of potentially missed cases over the last 12 months 3. THREE NEW AGENTIC MONITORING TRIGGERS For each proposed trigger: - Trigger name and natural-language description - Specific data inputs required (transaction fields, counterparty fields, trade-document fields) - Threshold logic with starting values and tuning guidance - Estimated alert volume per month at the starting threshold - False-positive rate target and the calibration plan 4. PROSECUTION-READY EVIDENCE - For each new trigger: what evidence must be captured at alert time to support a MAS-ready SAR - Audit trail requirements (cryptographic receipt, decision rationale, time-to-disposition) - Workflow integration: how the alert routes to the investigation team 5. REGULATORY ENGAGEMENT - Recommended language for the next MAS engagement on the gap-closure plan - Timeline for full implementation and pilot results - Open questions for legal counsel review Be specific. Cite the FATF Singapore review section references. Where the firm's input is insufficient to assess a gap, flag the question. Do not hedge the recommendations.
Test it in Claude or another LLM
This prompt is built for the Luna agent inside deepidv, where it cross-references your Singapore entity's live AML controls against the May 2026 FATF review, maps trade-based money laundering (TBML) detection gaps, and proposes new agentic monitoring triggers. Before running it against your real controls catalog, you can dry-run the same gap analysis in any LLM with a synthetic controls sample to verify the report structure and rigor.
- 1
Paste the full prompt into Claude, ChatGPT, or Gemini, but replace the opening "Luna, cross-reference our current AML controls..." with a role instruction such as "Act as an AML compliance analyst specializing in TBML and FATF mutual-evaluation gap analysis for a Singapore-regulated entity."
- 2
Paste the synthetic sample data block (below) so the model has a controls catalog, monitoring rules, conversion rate, and anonymized SAR counts to assess against the FATF criteria.
- 3
Tell the model that where the sample input is insufficient to assess a finding it should flag the open question rather than inventing coverage, matching the prompt's own instruction.
- 4
Check the output: for THIS prompt good output is a gap map that tags each FATF finding covered/partial/gap with cited section references, names the top 3 TBML detection gaps with the missed trade-pattern signature and 12-month missed-case volume, proposes exactly 3 new agentic triggers each with data inputs, threshold logic with starting values, monthly alert-volume estimate, and a false-positive target, plus prosecution-ready evidence and audit-trail requirements and suggested MAS engagement language. Reject hedged recommendations or any finding asserted without a section reference.
- 5
Once the report structure and specificity look right, run the prompt live in the deepidv dashboard where Luna executes it against your actual Singapore AML controls and transaction-monitoring rule set instead of the synthetic sample.
Synthetic sample data to paste alongside the prompt
Fake test data, safe to share with any LLM. Swap in your own once the output looks right.
Synthetic AML controls sample (clearly fake, anonymized): Entity: ACME-TEST-SG Pte Ltd | controls catalog: 14 rules (sanctions screening, PEP, structuring) | TBML-specific rules: 1 (round-dollar invoice flag only) Txn monitoring: threshold alerts on >SGD 20,000 wires; no trade-document cross-check Investigation-to-prosecution conversion: 4% | open advisories responded: MAS-TEST-2025-07 TBML SAR filings last 12mo: 9 filed, 1 actioned (anonymized) | sample case: counterparty BETA-TEST-LTD, over-invoiced electronics, ITIN ending 0000
Pairs with on deepidv
Sources & further reading
FAQ
How do I do a FATF gap analysis for my Singapore AML program?
Start from the most recent FATF mutual evaluation of Singapore and map each finding to a specific control in your AML catalog, marking it covered, partial, or gap with evidence attached. An AI agent like Luna can run this cross-reference in minutes if you provide your controls catalog and monitoring rule set, but every gap rating should be reviewed by your MLRO before it reaches MAS.
What is trade-based money laundering (TBML) and why does the FATF Singapore review focus on it?
TBML disguises illicit funds as legitimate trade by manipulating invoices, shipment values, and counterparty structures, and Singapore's position as a global trade hub makes it a priority typology there. FATF assessments look at whether firms can detect trade-pattern signatures like over-invoicing or circular shipping routes, which is why this prompt asks for the specific rule that would have fired had it existed.
Related prompts
Run it with live verification data
These prompts work in any LLM. Inside the deepidv dashboard, Luna, Arbiter, and Arc run them against your real sessions, screening lists, and audit trails.
Book a Demo