TraceLayer

Billions of dollars move through financial systems every day, and suspicious transactions rarely tell the whole story on their own. A single payment may look harmless, while the larger network of accounts, transaction timing, and movement of funds reveals a potentially suspicious pattern.

We were inspired by a fundamental problem in financial crime detection: finding a suspicious transaction is only the beginning. Investigators also need to understand why it was flagged, where the money came from, which data supports the alert, and whether the evidence can be trusted.

Traditional detection systems can generate alerts, but investigating those alerts often requires analysts to jump between transaction records, spreadsheets, graph tools, policies, and case notes. Meanwhile, poor data quality, missing lineage, and opaque model predictions can make decisions slower and harder to audit.

We wanted to bring these pieces together in one platform. That idea became TraceLayer, a financial intelligence and data-governance layer built to make every alert explainable, every source traceable, and every investigative decision auditable.

What It Does

TraceLayer connects financial transaction intelligence with governed investigation workflows.

AI-Powered Risk Detection: Uses an XGBoost model to score potentially suspicious transactions and help prioritize investigations.

Transaction Graph Investigation: Connects accounts and transactions to help investigators trace the movement of funds, discover connected entities, and explore suspicious paths and cycles.

Context-Aware AI Investigation Copilot: Designed to let investigators ask natural-language questions about a flagged transaction using its source records, linked entities, graph evidence, model explanations, and relevant policy documents.

Data Quality and Lineage: Connects alerts back to their underlying records and surfaces data gaps or quality issues that could affect investigative conclusions.

Policy-Grounded Intelligence: Retrieves relevant policy guidance to help explain why a risk signal matters and what an investigator should examine next.

Auditable Case Management: Structures the investigation around evidence, explanations, analyst notes, and human-reviewed dispositions.

The goal is not to let an AI declare someone guilty. It is to help a human investigator move from an unexplained alert to a well-supported, reviewable decision.