In financial services, time is money and nowhere is that truer than in reporting. One global financial services company was managing more than €2.8 trillion in assets across 1,200+ funds and struggling to keep up with monthly regulatory and executive reporting.
Their legacy setup involved four disjointed stages, manual ingestion, Excel-based cleansing, disconnected dashboards and on-prem governance resulting in a four-day reporting cycle with a 12 % error rate. Teams spent days reconciling figures and answering questions that should have taken minutes.
The Challenge
The business needed a modern, cloud-native approach to automate data ingestion, unify logic and deliver analytics in near-real time, while ensuring every stakeholder could trust the numbers.
As part of a broader business process change to Client On-boarding, Investment Statements (by Investment Managers) were processed, extracting 55 data entities to supplement Client information. These fields varied in complexity and free-text structure between Investment Managers, requiring intelligent parsing and consistent modelling before the data could add value downstream.
The Approach
The engagement was structured as a three-week rapid-prototype sprint designed to deliver a production-ready AUM (Assets Under Management) mart in Snowflake. The roadmap was simple:
- Automate ingestion using a low-code ELT connector for near-real-time pulls from custodians, trading platforms, and internal systems.
- Model and transform data in Snowflake using dbt for version control, testing, and modular SQL.
- Leverage AI and Snowflake Cortex to parse handwritten and free-text statements, standardising 55+ entities from Investment Managers.
- Expose data via interactive analytics through Streamlit and Cortex Analyst for real-time insights.
- Govern and productise data through Snowflake’s secure marketplace model, ensuring trust and reusability.
The Execution
Over 21 days, the project team implemented the entire stack:
- Days 1–10: Data ingestion and schema design. Snowflake was configured as the central warehouse, and ELT pipelines were established to unify source data.
- Days 11–14: AI-driven parsing was trained on handwritten investment documents, transforming unstructured PDFs into clean, relational tables. Manual QA dropped from 4 days to 30 minutes.
- Days 15–17: Interactive dashboards were deployed using Streamlit and Cortex Analyst, giving portfolio managers real-time drill-downs by region, asset class, and period.
- Days 18–21: The AUM mart was published as an internal data product, complete with metadata and refresh cadences, and piloted across three business units.
The Results
| KPI | Before | After |
| Report turnaround | 4 days | 2 hours (99 % faster) |
| Data quality errors | 12 % | < 1 % |
| Active dashboard users | 0 | 45 (30 % of investment managers) |
| Teams using data products | 0 | 12 within a month |
| Annual operational cost | – 66 % |
Beyond the metrics, the firm now operates from a single source of truth, with governed, version-controlled data and automated lineage tracking. Regulatory audits that once required weeks of manual validation are now traceable in minutes.
Why It Matters
By combining Snowflake Cortex’s AI capabilities with modern data engineering practices, the firm not only accelerated reporting but transformed data into a reusable asset. Investment managers can now query, slice, and drill into AUM insights without waiting for IT intervention.
Key Learnings
- Governance from day one prevents future security and compliance gaps.
- Iterative data product releases encourage faster feedback and adoption.
- AI + ELT = automation, together they eliminate human bottlenecks and improve accuracy.
- Cross-functional demos keep stakeholders aligned and invested in the outcome.
Looking Ahead
The next initiatives include expanding the AUM mart to include net new investor data for better attribution, and building a forecasting model to support strategic asset allocation all within the same Snowflake Cortex ecosystem.
Final Thought:
What started as a compliance-driven need evolved into a business-enabling data platform, delivering insights in hours instead of days. This use case demonstrates how Snowflake Cortex and modern ELT frameworks can re-define operational agility in financial services.
If your organisation is still wrestling with slow reporting, inconsistent data, or siloed analytics, a Snowflake-centric rebuild might be the fastest route to clarity, accuracy, and confidence in your numbers.
Interested in exploring how Snowflake Cortex can transform your reporting and analytics?
Contact our head of AI Jai Parmar : jai.parmar@leit.ltd