Financial crime that hides in the relationships.

The placement, layering, and integration chain is invisible to any rule that examines one transaction at a time. Purple8 traverses the full entity network at 4, 6, or 8 hops and traces the path from source funds to OFAC SDN in milliseconds. A chain that would take an investigator two weeks to reconstruct manually is a single graph query.

Two-tier SDN screening runs an offline local pattern match first, then an optional real-time API call to ofac.io, Dow Jones Watchlist, Refinitiv World-Check, or ComplyAdvantage. Every screened entity is a graph node with full relationship context. A name that clears direct screening but sits two hops from a sanctioned entity is still flagged, because the relationship graph catches what name-matching misses.

Fraud rings share devices, addresses, phone numbers, and IP addresses across seemingly unrelated accounts. Community detection clusters accounts that share any attribute node. PageRank identifies the most influential accounts in the network. Betweenness centrality finds the key intermediaries. A synthetic identity ring of 40 accounts spanning 3 years surfaces as a single connected component.

Multi-hop traversal maps the full ownership chain through nominee directors, Cayman LPs, BVI shells, and Singapore SPVs to the ultimate beneficial owner. DocIntel ingests company registry documents, KYC forms, and adverse media references and emits entities directly to the graph. A 5-entity offshore structure resolves to its UBO in a single traversal call.

A new person appears, takes Power of Attorney, and starts moving money. Purple8's Change Data Capture fires the moment a new CONTROLS or POA_FOR edge is written. Link prediction scores new beneficiaries with no prior relationship as anomalous. Community detection surfaces one person controlling multiple elderly customers across the book, a pattern that no per-account rule can see.

Multiple sub-$10,000 cash deposits by the same individual, related accounts, or shared devices constitute structuring, a federal crime. Purple8 aggregates across accounts linked by shared device, address, or relationship edge. The aggregated pattern that evades the CTR threshold surfaces as a typology match against the structuring pattern graph.

Over- and under-invoicing between related trading entities is the largest money laundering typology by volume and the hardest to detect. Purple8 connects the trade finance graph, covering invoices, letters of credit, counterparties, and jurisdictions, to the corporate ownership graph. When buyer and seller share a beneficial owner, the relationship edge makes the manipulation visible.

Spoofing, layering, bear raids, wash trading, and dividend arbitrage leave traces in the order graph. Purple8 models positions, orders, cancellations, and beneficial ownership as a unified graph. Two hedge funds shorting the same illiquid mid-cap on the same days via a shared prime account is a community detection result, not a manually configured alert rule.

The SAR filing workflow runs as a Journey Engine state machine: alert, triage, investigation, supervisor review, FIU escalation, narrative generation, filing. Each stage transition is a mandatory human decision recorded as an immutable graph edge. The SAR narrative is generated from the actual graph evidence, including the relationship chain, transaction timeline, and typology match, not reconstructed from memory after the fact.

The same individual appearing in Retail Banking, Asset Management, and Prime Brokerage with different risk profiles is the most dangerous customer in the book. Purple8's multi-tenant architecture keeps business line data isolated for regulatory purposes while letting Group Financial Crime query across all tenants. Cross-divisional alerts are graph edges, not email threads.