AWS re:Invent 2025 - LSEG Risk Intelligence: Accelerating World-Check with AI (GBL207)

Accelerating World-Check with AI: Transforming Financial Crime Detection

Addressing the Challenge of Financial Crime

• Financial crime, including fraud, money laundering, and terrorism financing, is a global problem costing trillions of dollars annually.
• Financial institutions struggle to keep up with the speed and sophistication of financial criminals, with 78% admitting they lack the necessary people and technology.
• LSEG Risk Intelligence, a trusted partner in financial crime detection, is on a mission to accelerate its World-Check platform using AI and automation.

LSEG Risk Intelligence: Powering Financial Crime Detection

• LSEG Risk Intelligence provides screening services, due diligence, account verification, identity verification, and onboarding solutions to over 10,000 customers globally.
• Their key focus areas are customer and third-party risk, as well as digital identity and fraud.
• The company aims to increase efficiency, leverage SaaS delivery, consolidate vendor complexity, and deepen insights through the responsible adoption of AI.

Accelerating World-Check with AI: A Phased Approach

1. Prompt-Only: Start with simple AI-powered actions to accelerate specific tasks, validating results with human oversight.

   • Example: Summarize a news article, extract key names and facts, and present the information to analysts for review.
   • Benefits: Quickly prove value, reduce risk, and build trust in the organization.

2. Single Agent Action: Implement AI agents to perform specific tasks within the curation workflow.

   • Example: Classify article relevance, extract entities, and propose a draft record.
   • Benefits: Accelerate individual steps in the process while maintaining human oversight.

3. Retrieval-Augmented Generation: Leverage large language models and vector databases to enrich and cross-reference data.

   • Example: Cluster related news articles, extract additional details about individuals, and provide a more comprehensive record.
   • Benefits: Improve data quality and insights by grounding AI outputs in real-world data.

4. Genetic Orchestration: Coordinate multiple AI agents within a defined workflow, with human-in-the-loop validation.

   • Example: Automate the end-to-end curation process, with agents performing tasks like summarization, extraction, and quality assurance, while allowing for human review and intervention.
   • Benefits: Streamline the overall process, maintain human oversight, and ensure regulatory compliance.

5. Interoperability: Leverage technologies like AWS MPC to enable seamless integration and orchestration of AI components.

   • Example: Build a modular, scalable platform that can easily incorporate new AI capabilities as they become available.
   • Benefits: Enhance flexibility, scalability, and the ability to adapt to evolving business needs.

6. Model Optimization: Fine-tune and optimize AI models for specific use cases, balancing performance, cost, and latency.

   • Example: Customize language models to better understand financial crime-related terminology and context.
   • Benefits: Improve the accuracy and relevance of AI outputs while managing operational costs.

Driving Successful AI Adoption

• Start with a "painful" workflow and focus on delivering a working solution in production, rather than grand transformation plans.
• Maintain a human-in-the-loop approach, leveraging AI to augment experts rather than replace them.
• Follow a maturity roadmap, gradually increasing AI capabilities as trust and confidence in the technology grows.
• Ensure AI is grounded in trusted data and maintains human oversight, especially in regulated industries like financial services.

Outcomes and Impact

• Reduced update times from hours to minutes, while maintaining accuracy and quality.
• Increased efficiency, with analysts freed up to focus on higher-value tasks.
• Scalable content capacity without proportional headcount growth.
• Earlier detection of content issues, building greater trust with clients.

Key Takeaways

1. AI augments experts, it does not replace them. Human expertise remains at the core of LSEG's risk intelligence services.
2. Successful AI adoption follows a maturity roadmap, starting with simple use cases and gradually increasing capabilities.
3. In regulated industries, AI must be grounded in trusted data and maintain human oversight to ensure compliance and trust.
4. The goal is evolution, not disruption - combining AI capabilities with domain expertise to scale impact without compromising trust.