Risk scoring and underwriting
Refresh credit and policy risk scores with explainable features, segment checks, and clear referral thresholds. Your underwriters see why a case moved.
Impact focus: better decisions with an audit trail
Financial services AI Singapore
AI that guards your bottom line and your regulatory standing
Your risk models may still reflect pre-pandemic behaviour. Your fraud alerts may be creating more review work than useful leads. Lorong AI builds AI risk scoring and fraud detection solutions around Singapore's regulated financial environment.
A confidential review of model performance, controls, and documentation.
Risk signals become decisions your compliance team can explain.
Model performance and risk scoring
Fraud detection and AML workflows
MAS-ready governance records
Built for regulated teams
Core AI capabilities for modern financial services
Start with one operational problem. Keep the controls visible as the system grows. Each capability fits existing data, review, and escalation processes.
Fraud detection and AML
Combine transaction behaviour, entity links, and investigator feedback to reduce false positives while preserving suspicious activity escalation.
Case result: 65% fewer false positives
Claims processing automation
Sort incoming claims, identify missing evidence, and route low-complexity cases for faster adjudication. Human review stays in the right places.
Insurance result: 55% shorter processing time
Portfolio risk analytics
Monitor concentration, early warning signals, and changes in financial statements across an SME portfolio before exposure becomes a surprise.
Decision support: earlier portfolio intervention
Field result
How a fintech scaled fraud defences without scaling the compliance team
A Singapore remittance fintech had a high alert volume and a small investigations team. Lorong AI connected real-time transaction graph analysis to its existing AML workflow, then added reason codes that investigators could use in suspicious activity reports.
1
Map the signalsLink accounts, devices, beneficiaries, and transaction patterns.
2
Score in real timePrioritise unusual relationships instead of sending every deviation to review.
3
Explain each alertGive the compliance team evidence they can challenge, record, and report.
Transaction data
Entity links
Reason codes
AML review
Risk
engine
engine
65%fewer false positives
98%true positive rate maintained
"The team gave our investigators a clearer reason for every alert. Our review queue became manageable without weakening our controls."Ragulan Coh, Chief Compliance Officer, Singapore remittance fintech
AI governance for banks and fintechs
Governance documentation that supports MAS scrutiny
Your documentation should help a reviewer understand the model quickly. Ours follows the FEAT pillars and maps controls to the decision being made.
Fairness
Test outcomes across relevant customer segments and record the thresholds used to assess material differences.
Compliance tip: document exclusion rules.
Ethics
Set boundaries for sensitive data, human oversight, customer impact, and acceptable model use before deployment.
Compliance tip: name the accountable owner.
Accountability
Assign responsibilities for approval, monitoring, incident response, change control, and retirement.
Compliance tip: keep a decision register.
Transparency
Make model purpose, inputs, limits, review steps, and customer communication clear to the people who rely on them.
Compliance tip: preserve reason codes.
Insurance claims AI
Turn claims from a cost centre into a customer retention engine
Claims teams lose time when every case follows the same path. AI can triage incoming evidence, flag fraud indicators during assessment, and route low-complexity claims for automated adjudication.
- Classify claims by complexity and urgency
- Spot inconsistent statements and supporting evidence
- Keep a human approval step for sensitive decisions
55%
shorter claims processing time
Triage handles routine cases early, while specialists receive the claims that need judgement.
Confidential starting point
Let's stress-test your current AI risk models before the regulator does
Tell us where the pressure sits. We will arrange a focused call and return a written model review summary covering performance, explainability, and governance gaps.