Anti-Money Laundering Intelligence
Quantum-inspired detection across six independent dimensions. Purpose-built for financial institutions that need answers no conventional AML tool can provide.
The Problem
Rule-based systems and supervised models are trained on known laundering patterns — structured deposits, threshold evasion, textbook network topologies. Professional launderers actively avoid these patterns. The result: your "clean" transaction population contains hidden laundering that your current tools were never designed to find.
Genuinely legitimate activity and sophisticated laundering are indistinguishable to rule-based checks. You don't know which is which.
Professional operations use several techniques simultaneously — amount structuring AND rate manipulation AND network layering. Single-dimension detection catches at most one.
No existing AML tool answers: are hundreds of flagged accounts one organization or hundreds of independent actors? Getting this wrong means missing the kingpin — or flooding FinCEN.
Our Approach
A single detection method catches a single laundering signature. We deploy six — each fully independent, each detecting a different operational signature.
Anomalous amount distributions, threshold gaming, systematic rounding, repeated micro-transactions. Validated at 99.1% against labeled laundering.
Hub-and-spoke topologies, regimented recipient patterns, institutional concentration. Identifies organizational infrastructure.
Fabricated exchange rates (10,000×+ off market), paired conversion paths, one-way crypto channels. Detects coordinated multi-currency operations.
Entities operating at multiple conversion rate points simultaneously. Flags the most sophisticated actors.
Cross-method clustering. Answers the question no other tool can: are these connected entities one organization?
Band-pass FFT on transaction timing. Detects scripted/scheduled automation — the only method that analyzes when transactions occur. Zero overlap with all other dimensions.
Each dimension is fully independent. When multiple dimensions flag the same entity, it's not redundancy — it's convergent evidence. Statistically, the overlap between independent detection dimensions is orders of magnitude higher than chance. Temporal FFT has zero overlap with entity groups, hubs, or FX rankings — it finds completely new targets.
The Investigator's Framework
| Question | What most AML tools provide | The gap | QuantumAML.International |
|---|---|---|---|
| Who do I investigate first? | A list of flagged accounts, sorted by a single score | One detection angle — multi-method threats may rank below single-signal noise | A ranked priority table integrating all six detection dimensions. Actionable, auditable order. |
| What kind of laundering is this? | A generic "suspicious activity" label | No operational intelligence — you don't know how they are laundering | Automatic typology classification: hub-spoke distribution, round-trip crypto-fiat, one-way channels, multi-channel layering. |
| Are these people connected? | Not answered | Nobody answers this. Getting it wrong means missing the kingpin or flooding FinCEN with redundant filings. | Operational groups identified through cross-method clustering. One organization or hundreds of independent actors? We tell you. |
| Are they running scripts? | Not answered | No AML tool analyzes when transactions occur. Automated laundering scripts leave a spectral fingerprint invisible to all other methods. | Temporal FFT analysis detects scripted automation — unnatural transaction rhythms, machine-precise timing, calendar-synchronized layering cycles. We tell you. |
Verticals
Money laundering adapts to the financial service it exploits. Our platform adapts accordingly. Banking is operational today. Gambling, securities, and exchange services are in development — built on the same quantum-inspired detection architecture.
Transaction monitoring, correspondent banking, trade finance, and retail AML. Six detection dimensions optimized for deposit-taking institutions.
Chip washing, player fund layering, affiliate payment laundering. Detection adapted to gaming-specific transaction flows.
Wash trading detection, pump-and-dump fund flows, cross-market manipulation patterns.
Forex bureau monitoring, remittance corridor analysis, cash-to-crypto conversion detection.
Validation
| Labeled laundering accounts | 222,522 |
| Correctly identified | 220,531 |
| Match rate | 99.1% |
| Missed | 0.9% |
Important: The 0.9% not classified as highest-risk are not missed — they are captured by the secondary and tertiary detection dimensions.
IBM Transactions for AML — HI-Large Accounts. ~180 Million transactions (17 GB raw), 2.1 Million accounts, 222 Thousand labeled records, spanning 97 days. Kaggle
When two independent detection dimensions flag the same entity, the probability of coincidence is effectively zero (p ≈ 0). Observed overlap is orders of magnitude higher than what independence would predict.
This is mathematical proof — not a tuning artifact — that the methods detect the same laundering operations through completely different signatures.
The Breakthrough
This is the capability that separates this platform from every other AML tool on the market — proven on the same IBM® dataset that delivered 99.1% detection accuracy.
An investigator receives a list of flagged accounts. Four hundred of them operate in US Dollars, with similar patterns. Are they four hundred independent launderers who happen to use similar methods — or one organization running four hundred accounts?
Four hundred cases. Four hundred SARs. Months of parallel investigations — or missed connections entirely.
In the IBM® benchmark, the platform clustered those 400 FX-anomalous entities into eight operational cells based on shared features across all six detection dimensions — currency infrastructure, counterparty networks, operational tempo, amount patterns, rate manipulation signatures, and temporal fingerprints.
Eight SARs — one per cell — naming the controller of each and connecting them through shared infrastructure evidence.
This is not an incremental improvement. It's a different kind of investigation.
Quantum Technology
The platform uses quantum-inspired classical algorithms — a quantum simulation framework that executes genuine quantum mathematics on standard computing infrastructure. No physical quantum computer, no cryogenics, no cloud quantum services required.
The platform applies quantum mathematical machinery at multiple phases:
Entities are not assigned to a single risk category. Each entity exists across multiple risk states simultaneously, revealing ambiguity that rigid classification would hide.
Accounts that appear unrelated in classical analysis exhibit quantum correlations when their transaction patterns are analyzed as interacting states rather than isolated records.
When one laundering pattern is confirmed, the quantum framework redistributes probabilities across the entire entity network. Legitimate patterns reinforce constructively. Laundering patterns produce destructive interference.
The platform computes values with no classical equivalent — complex probability amplitudes that reveal hidden anti-patterns. These are mathematically impossible in classical statistics but are a natural consequence of quantum state representation.
Why it's deployable today: The quantum simulation runs on our protected infrastructure. There is no hardware barrier between your compliance team and quantum-powered detection. This is quantum computing as a practical tool — not a laboratory experiment.
6th Detection Dimension
This is a novel methodology with no equivalent in any existing AML system. While all other methods analyze what amounts are transacted and who receives them, temporal analysis asks: when do transactions occur — and with what rhythm?
Accounts transacting precisely every 6.7 hours, around the clock. No human maintains this rhythm.
Morning and evening execution, consistent to the minute. Scripted twice-daily layering.
Slightly faster than 24h — the signature of a server-side cron job, not a human calendar.
Same day every week, same hour. Automated layering synchronized to a business calendar.
When two apparently independent accounts share the same 6.7-hour cycle with sub-second precision, they are not independent. They are controlled by the same automation infrastructure. Temporal FFT provides forensic evidence of common control — evidence that no amount-based or network-based method can produce. This is the fourth question every investigator should now ask: Are they running scripts?
Deployment
Analysis runs on our protected infrastructure — purpose-built workstation hardware with ECC memory, the financial industry standard for computation integrity. No software is installed on your systems. No cloud dependencies. No API calls to external services.
Standard transaction data — sender, receiver, amounts, currencies, dates. No pre-processing. No labeling. No configuration.
A complete investigation-ready report: executive summary, ranked priorities, typology classification, entity grouping, hub profiles, and statistical appendices — suitable for regulators, prosecutors, and board review.
The Report
The platform takes your transaction data and produces a complete investigative report — automatically. No manual analysis. No spreadsheet work. No waiting for a data science team.
The report includes:
No labeled training data
No pre-defined laundering patterns or rules
No manual feature engineering
No data science or machine learning expertise
No cloud infrastructure or API keys
The platform operates on standard transaction data — the same data your existing systems already produce. Analysis runs on our protected infrastructure under NDA, with no external dependencies.
The platform is designed for auditors and compliance teams — not data scientists. The output is a document you can put in front of a regulator, a prosecutor, or a board of directors.
📋 For evaluators, financial institutions, and government AML divisions: The complete AML investigation report (aml_investigation_report_v2.md) — generated automatically by the quantum pipeline — is available for qualified review upon request. Request a copy.
Why Now
Regulatory expectations are rising. FinCEN, the FCA, and EU regulators are increasingly asking not just "did you file SARs?" but "did you understand the organizational structure behind the activity you flagged?"
The old model — flag individual accounts, file individual SARs, move on — is no longer sufficient. Regulators want to see that you investigated connectivity. That you asked whether fifty flagged accounts are fifty problems or one problem with fifty accounts.
Next Steps
A technical demonstration on your data — or a representative sample — is the most effective way to evaluate the platform. The output is a complete investigative report you can review with your compliance team. Everything is treated under NDA.
Request a Briefing →