Customer case study
How AI agents are accelerating hit list review for compliance teams
Faster false positive resolution, clearer analyst reasoning, and a full audit trail without adding headcount.
Profile
- Industry: Investment management
- Location: Singapore (MAS-regulated)
- Entity type: Licensed fund manager
- Key challenge: Manual hit list review
- Solution: Prudexis AI Compliance Agent
- Outcome: Faster, auditable hit decisions
Background
Compliance teams at MAS-regulated fund managers must screen investors against global watchlists on an ongoing basis. When a new hit is recorded, analysts need to determine whether the potential match is genuine or a false positive.
For most teams this workflow is manual: pulling investor records, cross-referencing watchlist entries, evaluating name similarity, geography, and context, and then documenting reasoning.
Prudexis deployed an AI compliance agent directly in this process so analysts can ask questions in plain language and receive structured, reasoned assessments in seconds.
The challenge
Before introducing AI agents into hit review, the team faced three recurring constraints:
- High false positive volume: screening generated large numbers of low-quality matches that still required human verification.
- Slow case-by-case investigation: analysts spent hours retrieving records, interpreting watchlist entries, and writing rationale for hits that were clearly unrelated.
- Inconsistent documentation: decision logic varied by analyst and format, weakening governance consistency and audit readiness.
The solution
Prudexis introduced an AI compliance agent that analysts query directly about open hits. The agent retrieves investor profile data, cross-references watchlist records, and returns a structured assessment with recommendation and rationale.
The interaction is conversational: analysts ask in plain language and receive a decision-ready report they can review and approve.
Live agent interaction - hit review
ANALYST
Hi agent, good morning — can you help me look at Aiden Tan? I saw there are new hits recorded. I need to understand whether you think these are false or positive.
PRUDEXIS AGENT
Analysis for Aiden Tan (Customer ID: 1000000003)
I've reviewed the screening hit. Here is my assessment.
Investor profile: Malaysian national, DOB 1981-01-07, residing at 18 Jalan Ampang, Kuala Lumpur. Risk level: Low. Account status: Inactive.
Screening hit: Watchlist match against KLARISSA NICOLE WHITE — US SAM Procurement Exclusions, Killeen TX 76541, USA. Match score: 78.32. Match reason: Name similarity.
✓ Recommendation: False Positive
Rationale:
- Complete name mismatch — "Aiden Tan" shares no meaningful similarity with "KLARISSA NICOLE WHITE"
- Geographic disconnect — Malaysian national in Kuala Lumpur vs. Texas, USA; no apparent connection
- Questionable algorithm match — the 78.32 score appears to be a technical artefact with no logical basis
- Context mismatch — US SAM Exclusions relate to federal procurement; unlikely to be relevant to this investor's profile
Suggested action: Close and dismiss this hit as a false positive, with notation that there is no reasonable connection between the investor and the watchlist entry.
The agent surfaces the right data, applies consistent analytical criteria, and produces a documented rationale — all within a single interaction. The analyst reviews, approves, and moves on.
The results
- 70% reduction in time spent per hit review
- 100% structured rationale captured for every decision
- Same-day hit queue clearance instead of multi-day backlog
Analysts now review agent-generated assessments instead of building them from scratch. Documentation quality improved as a side effect because each decision is packaged with a consistent rationale.
"Our analysts were spending most of their time on hits that were obviously wrong. The agent handles triage so we focus on cases that need human judgement."
Head of Compliance, MAS-licensed fund manager, Singapore
Why it worked
- Contextual reasoning: the agent explains why a match is or is not meaningful using profile, geography, source, and watchlist context together.
- Consistent output format: each assessment uses the same structure for easier review, approval, and archiving.
- Natural language interface: analysts ask questions the way they already think and work.
- Integrated data access: the agent pulls from Prudexis investor and watchlist data already in the platform.
Regulatory responsibility
Prudexis is compliance software and does not replace your internal AML/CFT programme, legal advice, or licensing obligations. AI agent outputs are advisory. Each firm remains responsible for final case decisions, policy interpretation, and local regulatory filings. Prudexis supports implementation with clear controls, structured documentation, and a full audit trail.
See the agent in your workflow
Book a demo to see how Prudexis handles hit review, false positive triage, and compliance documentation in one platform.