AML Compliance Infrastructure · Built for Brazilian Financial Institutions

Auditable
intelligence

The AML platform that replaces black-box scores with legally defensible evidence.
Every alert mapped to BACEN 3.978 and COAF typologies, generated with its own audit trail.

Get in touch contact@nllabs.ai Investors & Funds funds@nllabs.ai
Scroll
01 — Problem

Replacing
opaque scores

The market doesn't need another generic risk score. It needs auditable, legally defensible compliance trails.

Legacy AML systems issue a number and leave the burden of explanation on the compliance desk. We invert that — every alert arrives with its evidentiary chain already attached.

Status Quo
The Glass Box
Black-box machine learning models.
Explainable AI models with traceable logic.
Generic risk scores (0–100) without context.
Automated legal foundation citing BACEN & COAF typologies.
Disconnected silos requiring manual legal review.
Interconnected corporate graph networks mapped in real-time.
02 — Framework

Four pillars
of trust

Explainability
Every alert is accompanied by a human-readable justification. No black-box decisions.
XAI · SHAP / LIME
Control
Strict governance over data, ensuring absolute compliance with LGPD and strict CNJ cloud restrictions.
LGPD · CNJ Compliant
Auditability
A completely traceable, mathematically proven path from the raw data source to the final ML decision.
Traceable · End-to-End
Responsibility
Aligning AI outputs directly with legal frameworks, including Law 9.613/1998 and Circular BACEN 3.978/2020.
Law 9.613 · BACEN 3.978
03 — Architecture

Five-layer
multi-agent stack

Each layer builds on the last. Detection surfaces the anomaly. Graph AI reveals the hidden network. The Legal LLM maps it to specific BACEN and COAF typologies. XAI produces statistical proof. Real-time monitoring pushes it to the operator.

L / 05
Real-Time MonitoringKafka-like streaming pipeline for continuous transactional processing and urgent alerting.
Streaming · Alerts
L / 04
Explainable AISHAP / LIME integration providing statistical proof alongside the generated legal text.
SHAP · LIME
L / 03
Legal AILLM that generates exact legal reasoning, mapping anomalies directly to COAF / BACEN norms.
LLM · Typology Map
L / 02
Graph AIReal-time structural mapping of corporate networks and hidden financial triangulations.
Graph · Networks
L / 01
Detection CoreXGBoost and Isolation Forests for anomaly detection — smurfing, layering, structuring.
XGBoost · Isolation Forest
"Suspicious operation framed in COAF Typology X (fractioning), per Circular BACEN 3.978/2020. Communication within 24h recommended."
04 — Output

A diagnostic
that explains itself

This is what reaches the compliance analyst — not a number in a vacuum, but a complete legal rationale with citations, risk factors, and the statistical proof behind each flag.

Entity Diagnostic · Exhibit A Live · 10:05:32
Overall Risk Score
85/100
Classification: High
Identified Risk Factors
✓
Passive party in an active judicial process
DataJud
✓
Classified as Politically Exposed Person
PEP
✓
Partner in companies with high-risk AML CNAEs
Receita Federal
✓
Hidden corporate links to laundering fronts
Graph AI
Orange Profile Indicator · 56.05 Law 9.613/1998 · BACEN 3.978/2020
05 — Data

The regulatory
ecosystem

Target state: 80–100 databases. Phase 1 delivers the core public sources — judicial, corporate, sanctions, governmental — fused into a single auditable graph.

Set 01 — Regulatory BACEN & COAF
  • BACEN (SCR / CCS)
  • COAF typologies
  • Open Finance behavioral
Set 02 — Identity Corporate Registries
  • Receita Federal (QSA)
  • B3 public data
  • PEP lists (electoral, donations)
  • Corporate family trees
Set 03 — Judicial Courts & Sanctions
  • DataJud (CNJ)
  • Interpol & OFAC
  • CEIS / CNEP / CEPIM
  • National Slave Labor lists
Set 04 — Governmental Federal Transparency
  • Transparency Portal
  • PNCP (Public Contracting)
  • Transferegov
06 — Audience

Built for
major institutions

nl.ai is designed for the largest financial players in the market — institutions whose compliance operations demand not just detection, but legally defensible evidence at scale.

Segment 01
Tier-1 retail & commercial banks
High-volume transactional environments where every alert must withstand regulatory scrutiny and internal audit.
Segment 02
Private banking & wealth management
PEP-heavy portfolios and complex corporate structures that expose institutions to heightened AML liability.
Segment 03
Asset managers & fintechs at scale
Large-scale operations required to communicate with COAF and defend automated decisions under BACEN supervision.
Segment 04
Regulated exchanges & payment institutions
High-throughput infrastructures where black-box scoring is no longer sufficient to meet compliance obligations.

Auditable evidence,
not just alerts

We do not merely detect suspicious operations. We understand the risk legally, justify every decision using regulatory norms, and deliver instantly auditable evidence.

If you're building or evaluating an AML stack and care about defensibility, we'd like to talk.

Who you areIntroduce yourself and your role
InstitutionYour bank, wealth firm, or regulator
Use caseWhat you're trying to monitor or defend
TimelineWhen you need to move