How Complii Really Works: A Guide for Compliance Officers and AML Specialists

Key Benefits of UCID:

• ✓Unified Risk View: All risk signals, screening results, and transaction patterns are aggregated at the UCID level, providing a complete picture of customer risk regardless of how many customer records exist.
• ✓Eliminates Duplicate Work: Screening and risk evaluation run once per identity cluster, not per customer record, reducing operational costs and preventing redundant alerts.
• ✓Consistent Monitoring: Transaction monitoring scenarios evaluate behavior at the UCID level, ensuring that structuring or other suspicious patterns are detected even when transactions are spread across multiple customer records.
• ✓Non-Destructive: UCID linking preserves all original customer records and their audit trails, unlike traditional merge operations that can lose historical data.

Advanced Matching Algorithms:

• SimHash Fingerprinting: Converts names into 64-bit fingerprints where similar names differ by only a few bits. This enables fast comparison and handles minor variations effectively.
• Character Trigrams: Breaks names into three-character sequences to detect typographical errors and misspellings. If one letter changes, many trigrams still overlap, maintaining match detection.
• Phonetic Matching (Soundex): Handles names that sound similar but are spelled differently, addressing variations like 'Mohammad' vs 'Mohamed' or 'Smith' vs 'Smyth'.
• Jaro-Winkler & Monge-Elkan: String similarity algorithms that handle token reordering, missing tokens, and mild misspellings. These provide the final similarity score (0-100%) that determines if a match is returned.

What you get

• ✓Explainable evidence links back to the source URL (with extracted source domain).
• ✓Filters to reduce noise: minimum sources/articles and optional state-actor/entity-type exclusion.
• ✓UCID matching so the signal attaches to the identity cluster, not a single customer record.

Risk Categories:

• Document Risk (max 30 points): Expired documents, missing required documents, document mismatches, or poor document quality.
• Identity Risk (max 30 points): Screening matches against sanctions, PEP lists, or custom block lists. Higher scores for confirmed matches vs. potential matches.
• Geographic Risk (max 20 points): Customer location, transaction corridors, or addresses in high-risk countries as defined by your country risk configuration.
• Network Risk (max 20 points): Shared phone numbers, email addresses, or physical addresses with other high-risk customers. Indicates potential money laundering networks.
• Behavioral Risk (max 10 points): Unusual transaction patterns, odd-hour transactions, or deviations from expected behavior profiles.
• External Risk (max 40 points): Custom risk signals from external systems, manual risk adjustments, or other tenant-specific risk factors.

Monitoring Scenario Types:

• Threshold Monitoring (tm001_sum): Detects when the rolling sum of transaction amounts exceeds a configured threshold within a time window (e.g., $50,000 in 7 days). Useful for identifying large-value transactions that may require reporting.
• Velocity Monitoring (tm002_count): Counts the number of transactions within a time window and alerts when the count exceeds a threshold (e.g., 20 transactions in 1 day). Identifies rapid transaction patterns that may indicate structuring.
• Structuring Detection (tm003_structuring): Specifically designed to detect transactions that are just below reporting thresholds (e.g., multiple $9,500 transactions when the threshold is $10,000). This is a key indicator of money laundering attempts.
• Behavioral Anomaly Detection (tm004_behavioral): Compares current transaction patterns against a historical baseline. Alerts when current activity deviates significantly from normal behavior (e.g., 2x increase in transaction volume).
• Geographic Monitoring: Filters transactions by specific corridors (country pairs) or funding institutions. Useful for monitoring high-risk corridors or specific payment channels.

AI Capabilities:

• AI Document Analysis: Extracts customer data from document images (passports, ID cards, driver's licenses) using OpenAI Vision. Automatically populates customer fields including name, date of birth, nationality, address, and document numbers with high accuracy. Supports multiple document types and languages.
• AI Customer Creation: Creates complete customer records from 1-2 document images. Simply upload photos of IDs or passports, and the AI extracts all necessary information and creates the customer automatically. This dramatically speeds up onboarding while reducing data entry errors.
• AI Screening Review: Automatically reviews screening alerts using GPT-4o. The AI analyzes customer information and screening matches, then makes binary decisions: Confirmed (true match) or False Positive. For confirmed matches, cases are automatically created. For false positives, risk scores are automatically recalculated. This feature can process alerts 24/7, ensuring rapid response times even outside business hours.

Case Workflow:

1. 1.Alert Creation: Screening matches or transaction monitoring alerts are created automatically. AI review can pre-classify alerts as Confirmed or False Positive.
2. 2.Case Creation: Analysts or the system (for AI-confirmed matches) create cases and link relevant alerts and evidence.
3. 3.Assignment & Review: Cases are assigned to analysts who review all linked evidence, add notes, and investigate further if needed.
4. 4.Disposition: Analysts set case disposition (True Match, False Positive, SAR/STR filing, Other) and add a summary documenting their decision.
5. 5.Closure: Once disposition is set and all documentation is complete, cases are closed. All actions are logged in the audit trail for regulatory review.

Network Analytics Capabilities:

• ✓Flow Patterns: Visualize transaction flows between identity clusters. Identify top senders, top recipients, and unusual concentration patterns.
• ✓Risk Corridor Analysis: Automatically identify high-risk transaction corridors based on country risk scores. Filter transactions by origin and destination countries.
• ✓Concentration Detection: Detect unusual concentration of transactions to/from specific parties. High concentration may indicate money laundering networks or structuring.
• ✓Evidence Linking: Click any network connection to view all transactions between two identity clusters, providing complete evidence for investigations.

Audit Features:

• ✓Complete Change History: Every customer update, screening run, risk score change, case status change, and configuration modification is logged with timestamps and user information.
• ✓Before/After Snapshots: For critical entities (customers, cases, alerts), the system stores JSON snapshots of data before and after changes, allowing you to see exactly what changed.
• ✓Event Logging: All case events (assignments, status changes, notes, dispositions) are logged in an append-only event log that cannot be modified, ensuring data integrity.
• ✓Export-Ready Reports: Generate compliance reports with all audit information for regulatory submissions. Reports include screening history, risk score evolution, case documentation, and transaction evidence.

Issue 1: High False Positive Rates

Traditional screening systems generate many false positives, overwhelming analysts and reducing trust in the system. Common names match everything, and analysts spend most of their time dismissing irrelevant alerts.

Complii solves this through configurable screening presets that balance precision and recall. The FP2 fingerprinting algorithm handles misspellings and name variations without generating noise for common names. AI screening review automatically classifies 70-80% of alerts as false positives, allowing analysts to focus on true risks. Document-based matching provides 100% accuracy when document numbers are available.

Issue 2: Fragmented Customer Identity

Customers are registered with slight name variations, different addresses, or multiple identifiers across systems. This leads to duplicate screening, fragmented risk profiles, and missed suspicious patterns that span multiple records.

UCID (Universal Customer Identity) clustering automatically groups related customer records into identity clusters. Screening, risk assessment, and transaction monitoring all operate at the UCID level, ensuring consistent risk evaluation and pattern detection regardless of how many customer records exist. The system maintains all original records and their audit trails, so no historical data is lost.

Issue 3: Manual Transaction Monitoring

Many organizations rely on manual review of transactions or simple rule-based systems that miss complex patterns. Structuring, layering, and behavioral anomalies go undetected, creating compliance gaps.

Complii provides automated transaction monitoring with multiple scenario types: threshold monitoring, velocity detection, structuring detection, behavioral anomaly detection, and geographic filtering. Scenarios are evaluated continuously at the UCID level, ensuring patterns are detected even when transactions span multiple customer records. Each alert includes a complete explainable payload for regulatory review.

Issue 4: Lack of Audit Trail

Regulators require complete audit trails showing who made decisions, when, and why. Many systems lack comprehensive logging, making regulatory reviews difficult and exposing organizations to compliance risk.

Complii is audit-ready by default. Every action is logged with timestamps, user information, and before/after snapshots. Case management provides complete documentation workflows with linked evidence. All screening runs, risk score changes, case dispositions, and configuration changes are traceable. Reports can be generated for regulatory submissions with complete audit information.

Issue 5: Slow Customer Onboarding

Manual data entry from documents is time-consuming and error-prone. Analysts spend hours typing information from passports and ID cards, delaying customer onboarding and increasing operational costs.

AI document analysis extracts customer data from document images automatically. Upload a photo of an ID or passport, and the system populates all customer fields with high accuracy. AI customer creation allows you to create complete customer records from 1-2 document images in seconds. This reduces onboarding time from hours to minutes while eliminating data entry errors.