A Risk and Research Management Software Company Uses AI to Uncover Hidden Risks in Global Supply Chains
At a glance
A risk and research management software company partnered with New Math Data to build an AI-powered research and discovery solution that surfaces hidden risks in global supply chains, including modern slavery, forced labor, and illicit criminal networks. The solution enables conversational access to structured and unstructured data, including spreadsheets, reports, and infographics, allowing analysts to surface insights and relationships in seconds.
Industry
Use Case
Intelligent document processing and AI-augmented enterprise search
Solution implemented
- Built an AI-assisted research platform powered by Amazon Bedrock and Claude 3.5 Sonnet for conversational natural-language querying
- Developed multimodal ingestion pipelines with AWS Lambda, Glue, and S3 to process structured (Excel, CSV) and unstructured (PDF, image) documents
- Indexed content into an Amazon Bedrock Knowledge Base with vector embeddings stored in Aurora PostgreSQL (pgvector) for semantic retrieval
- Deployed a Chainlit conversational interface hosted on EC2 for real-time, context-aware querying
- Provisioned and managed infrastructure via AWS CloudFormation for reproducibility and resilience
- Implemented full observability through CloudWatch and secured all components using IAM RBAC, KMS encryption, and VPC isolation
The value equation
- Faster, more accurate analysis of global risk and criminal-network data
- Improved data accessibility and discoverability through natural-language search
- Expanded ability to process multimodal documents, including charts and graphics
- Automated ingestion pipelines that scale to high data volumes
- Unique market differentiation for the company as an AI-driven leader in ethical supply-chain intelligence
Company Snapshot
A mission-driven investigative research company that combines advanced AI technology with human expertise to deliver risk intelligence to clients in more than 160 countries.
Location
Maryland, United States
AI Solution Makes Global Risk Intelligence Instantly Accessible
A risk and research management software company helps organizations identify and mitigate hidden risks within global supply chains. Its proprietary datasets and investigative reports span areas such as forced labor, corruption, and links to criminal enterprises. However, this wealth of intelligence was dispersed across multiple formats, making it difficult to search, analyze, and synthesize efficiently.
To address this, the company partnered with New Math Data to develop an AI-powered, retrieval-augmented generation (RAG) platform that unifies complex, multi-format data into an interactive, searchable knowledge base. The system enables investigators, compliance teams, and clients to explore their data conversationally—significantly improving accessibility and accelerating time-to-insight.
Problem
Analysts faced challenges navigating massive spreadsheets of interconnected data, reading thousands of global reports, and managing unstructured documents. These manual processes created a bottleneck between data collection and insight generation.
The company’s investigative teams and clients needed a scalable, production-ready solution that could:
- Automatically ingest and process structured and unstructured datasets from global sources
- Comprehend and index data from multiple formats, including spreadsheets, PDFs, and infographics
- Provide semantic and keyword-based search to reveal relationships across data types
- Enable non-technical users to query and interpret findings in natural language
- Ensure secure access controls for sensitive investigative data
Solution
AI-Powered Conversational Research Platform
New Math Data built an end-to-end AI research assistant that allows users to ask questions directly against the proprietary and external data sources. The system transforms multimodal information, such as text, tables, charts, and images, into a unified knowledge base.
Multimodal Ingestion and Transformation
Automated pipelines built on AWS Lambda, Glue, and S3 extract and process Excel, CSV, PDF, and image files. PDFs are converted into image sequences and analyzed using the multimodal model Claude 3.5 Sonnet, ensuring that both text and graphics contribute to the searchable dataset.
Dynamic Knowledge Base and Semantic Search
Processed content is indexed in an Amazon Bedrock Knowledge Base and embedded into Aurora PostgreSQL (pgvector) for high-precision semantic retrieval. This architecture supports deep contextual understanding and similarity-based search patterns.
Real-Time Conversational Interface
A Chainlit frontend hosted on EC2 provides a responsive chat-style interface where analysts can query the system in natural language. Each question triggers content retrieval from the Bedrock Knowledge Base, dynamically constructing context for accurate, multi-turn responses.
Infrastructure as Code and Observability
All infrastructure components, S3, Lambda, Glue, Aurora, EC2, and IAM roles, are managed via AWS CloudFormation, supporting full DevOps automation. CloudWatch monitors system health, performance, and API activity, with alerting and escalation for rapid issue resolution.
Security and Compliance by Design
The platform incorporates IAM role-based access, KMS encryption, and VPC isolation to safeguard data. These measures meet best practices for secure cloud architecture in regulated and high-sensitivity environments.
Over a focused six-week engagement, New Math Data delivered a production-ready, scalable AI platform that unites data integrity with accessibility, helping advance transparency, accountability, and ethical decision-making worldwide.
Related Case Studies
Vertically Integrated Utility Company
Migration of data systems and applications to AWS.
Ready to Transform Your Business?
See how New Math Data can transform your business with AWS-powered innovation.