[ts-evidence-1774692126261] Customer Identification

Customer Identification groups 1 use cases in finance around Financial Crime Compliance academic source 1. Query: "financial crime compliance" (AML OR "anti-money laundering" OR KYC OR sanctions OR fraud) AI benchmark OR evaluation OR precision recall site:arxiv.org OR site:openreview.net OR site:acm.org OR site:ieeexplore.ieee.org

The Problem

“Automate KYC/AML onboarding and monitoring with auditable AI-driven workflows”

Organizations face these key challenges:

Onboarding queues grow due to manual document review and repetitive screening steps

High false positives from watchlist/adverse media screening overwhelm analysts

Inconsistent risk ratings and case notes make audits and SAR/STR evidence hard

Ongoing monitoring (transactions/crypto) lacks timely escalation and clear rationale

Impact When Solved

Faster, streamlined customer onboardingReduced false positives by 70%Automated compliance documentation generation

The Shift

Before AI~85% Manual

Human Does

•Manual document review
•Identity verification
•Risk scoring
•Writing case narratives

Automation

•Basic alert generation
•Keyword matching for screening

With AI~75% Automated

Human Does

•Final approvals on complex cases
•Strategic oversight
•Addressing edge cases

AI Handles

•Entity resolution
•Document intelligence for verification
•Multi-step compliance orchestration
•Automated risk prioritization

Operating ModelHow It Works

How [ts-evidence-1774692126261] Customer Identification Operates in Practice

This is the business system being implemented: how work is routed, which decisions stay human, what gets automated, and how success is measured.

Operating Archetype

Optimize & Orchestrate

AI runs the engine. Humans govern.

AI Role

Operating Engine

Human Role

Governor

Authority Split

AI runs the workflow continuously; humans set policy and intervene on exceptions.

Operating Loop

This is the business workflow being implemented. The four solution levels are different ways to operationalize the same loop.

AIStep 1

Sense

Take in live demand, capacity, and constraint signals.

AIStep 2

Optimize

Continuously compute the best next allocation or action.

AIStep 3

Coordinate

Push those actions into systems, channels, or teams.

HumanStep 4

Govern

Humans set policies, objectives, and overrides.

AIStep 5

Execute

Run the approved operating loop continuously.

FeedbackStep 6

Measure

Measured outcomes feed back into the optimization loop.

Human Authority Boundary

The system must not approve or reject high-risk customers, beneficial ownership determinations, or suspicious activity decisions without human review.