CUSTOMER STORY
95%+ accuracy in AI-driven trade data extraction from term sheets
6-8 hours reduced to 15 minutes per trade processing time
80+ data products onboarded across Treasury and Finance operations
The World Bank Group is one of the world’s largest multilateral development institutions, working across 189 countries to reduce poverty and advance shared prosperity. Its Treasury division manages portfolio investment, trading operations, and financial services that support approximately 80 developing countries. With a global portfolio exceeding $100 billion, the Treasury depends on timely, reliable, and secure data to guide strategic decision-making, monitor global markets, and deliver financial services with consistency and integrity. As data volumes and complexity continued to grow, the organization recognized the need for a unified, scalable platform capable of powering modern analytics and emerging AI initiatives.
For years, the Treasury operated within a fragmented data landscape spanning more than 80 treasury and finance systems. This environment routinely produced conflicting financial metrics, undermining confidence in reports delivered to senior leadership and exposing longstanding data quality challenges. Without a unified architecture to reconcile disparate sources, teams struggled to maintain a single source of truth for mission-critical insights.
Manual workflows further limited operational efficiency. Traders spent six to eight hours reviewing term sheets, extracting key fields, and entering them into trading platforms—an error-prone process that slowed execution and diverted time away from high-value market analysis. Traditional warehouse-centric approaches added additional burden through redundant data copies, ongoing pipeline maintenance, and delayed access to new attributes or data sources.
Dependence on IT for reporting and data modeling also restricted agility. Analysts frequently waited for technical teams to prepare datasets or build new views, creating bottlenecks that hindered timely decision-making. As global financial markets accelerated, the lack of consistent, accessible, and governed data became a significant obstacle to the Treasury’s ability to operate efficiently and respond quickly to changing conditions.
The World Bank Treasury modernized its financial data infrastructure with Dremio, launching the “Finance One Lake” strategy. This initiative unified data from over 80 Treasury and Finance systems, replacing years of fragmentation. Deployed on Azure Kubernetes Service and connected to Azure Data Lake Storage, Dremio established a zero-ETL, schema-on-read architecture, eliminating data copying and enabling direct querying at the source.
A key component is a centralized semantic layer in Dremio, which ensures all analytics tools, including Power BI and Tableau, operate from consistent business logic, offering true plug-and-play flexibility. This unified layer also supports SHASTRA, an AI-powered trade automation system that uses Azure OpenAI for term-sheet data extraction and verified “golden copy” generation.
Governance and security are integrated using Azure Active Directory and Dremio’s identity-based controls, ensuring users across various workflows only access authorized data while enabling broad self-service. By standardizing on Parquet and open lakehouse principles, the Treasury created an open, scalable foundation ready to evolve with technologies like Apache Iceberg and Polaris, avoiding vendor lock-in and ensuring long-term flexibility for expanding analytics and AI initiatives.
The implementation leveraged Dremio's virtual datasets (VDS) and physical datasets (PDS) to create rapid iterations and prototyping capabilities, allowing data teams to quickly explore and prepare data before committing to formal ETL processes. Shell's teams utilized Dremio's reflections feature to optimize performance for frequently accessed datasets, though they learned to carefully manage and isolate these reflections as they became more heavily utilized across the organization.
Dremio's fine-grained access control capabilities proved essential for Shell's governance requirements, enabling secure provisioning of data spaces that could be safely assigned to Azure Active Directory groups. This approach maintained strong security standards while enabling self-service analytics across different teams and functions. The platform's sophisticated compute engine provided the performance needed to handle Shell's demanding workloads, including the ability to process billions of records within the tight timeframes required for production forecasting models.
The Dremio-powered lakehouse initiative has delivered significant and measurable impact across the World Bank Treasury’s operations. Trade processing, which once required six to eight hours of manual effort, now takes roughly fifteen minutes end-to-end, including human review. The AI extraction engine consistently achieves 95% accuracy, allowing traders to quickly validate only the remaining edge cases through a streamlined interface.
These efficiency gains have fundamentally reshaped trader productivity. Instead of spending the majority of their day manually parsing term sheets and entering data, traders can now focus on market analysis, pricing strategy, and opportunity identification—activities especially critical during periods of market volatility. This shift has strengthened the Treasury’s ability to respond quickly to changing conditions while improving the overall quality of decision-making.
The implementation of a single semantic layer within Dremio has also resolved long-standing data consistency issues. Teams that once produced conflicting financial metrics now rely on a shared source of truth, ensuring that leadership receives accurate, aligned, and timely reports. The platform has scaled rapidly, onboarding more than 80 data products and consolidating approximately 70% of the Treasury’s finance data.
Business agility has improved markedly as well. Analysts can now bring their own data, blend it with trusted corporate datasets, and perform advanced analysis without waiting for IT to build models or pipelines. These self-service capabilities operate within a robust security and governance framework, ensuring teams move quickly without compromising compliance.
Finally, the unified lakehouse foundation has positioned the Treasury for accelerated AI adoption. With clean, well-governed data now readily accessible, the organization is preparing new AI-driven capabilities, including natural language querying available in upcoming Dremio releases. Future enhancements will further strengthen the Treasury’s role as an innovator in global capital markets operations.
customer stories
