Compliance Prompt for AML Transaction Pattern Analysis
This review prompt turns pasted transaction data into a structured AML triage report covering structuring below reporting thresholds, velocity anomalies, FATF high-risk geographies, network clusters and behavioral deviations from the customer profile. Every finding lands in a table with a HIGH, MEDIUM or LOW risk rating and a recommended action, including when a SAR or STR filing should be considered. It is built for AML analysts, MLROs and fintech compliance teams who need a fast first pass over transaction exports.
How to use this prompt
- 1
Paste the prompt into your deepidv dashboard agent, Claude, ChatGPT or Gemini, then attach or paste your transaction export (CSV or table) as the next message.
- 2
Customize the reporting thresholds for your jurisdictions; the defaults cover the 10,000 dollar US and 15,000 euro EU levels but your local rules may differ.
- 3
Expect a structured findings table: account ID, pattern type, description, risk rating and recommended action across all five analysis dimensions.
- 4
Escalate HIGH findings to your MLRO for investigation and SAR/STR review; treat the output as analyst triage, never as an automated filing decision.
- 5
For continuous coverage, move recurring patterns into deepidv transaction monitoring rules so they fire in real time instead of in batch reviews.
The prompt
You are an AML transaction monitoring analyst. I will provide you with transaction data. Your job is to analyze the data for patterns that may indicate money laundering, terrorist financing, fraud, or sanctions evasion. For each dataset I provide, perform the following analysis: 1. STRUCTURING DETECTION - Identify transactions that appear designed to avoid reporting thresholds (e.g., multiple transactions just below $10,000 in the US, €15,000 in the EU, or other jurisdiction-specific thresholds) - Flag rapid sequential transactions to/from the same parties - Identify round-number patterns that suggest deliberate structuring 2. VELOCITY ANALYSIS - Flag accounts with sudden increases in transaction frequency or value - Identify deviations from the account's historical baseline - Note any accounts that went from dormant to highly active 3. GEOGRAPHIC RISK - Flag transactions involving high-risk jurisdictions (FATF grey list, black list, or sanctioned countries) - Identify unusual geographic patterns - Note cross-border patterns that may indicate layering 4. NETWORK ANALYSIS - Identify clusters of accounts that appear connected - Flag circular transaction patterns - Note any patterns consistent with funnel accounts or mule networks 5. BEHAVIORAL ANOMALIES - Flag transactions inconsistent with the customer's stated profile - Identify unusual timing patterns - Note any patterns that deviate from the peer group baseline For each finding, assign a risk rating: 🔴 HIGH (likely suspicious, recommend SAR/STR filing), 🟡 MEDIUM (warrants investigation), 🟢 LOW (unusual but likely explainable). Present findings in a structured table with: Account ID, Pattern Type, Description, Risk Rating, Recommended Action. I will now provide the transaction data.
Pairs with on deepidv
FAQ
Can AI detect money laundering patterns in transaction data?
An LLM with a structured analysis prompt can triage transaction exports for classic AML typologies such as structuring, layering, sudden velocity changes, funnel accounts and high-risk corridors, and rank findings by severity. It is a force multiplier for the first pass, but a human analyst must validate findings and make any SAR or STR filing decision.
What is structuring in AML and how is it detected?
Structuring is splitting transactions to stay just below mandatory reporting thresholds, such as the 10,000 dollar currency transaction reporting level in the US. It is detected by flagging clusters of near-threshold amounts, rapid sequential transfers between the same parties, and suspicious round-number patterns, all of which this prompt checks automatically.
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