APAC bank automates financial data analysis process: 90% cost reduction and 52x productivity gains with AWS-powered MLOps and GenAI
Corporate banking division partners with Amdocs to modernize manual financial analysis using AWS powered MLOps and GenAI, creating automated platform that dramatically improves speed, accuracy, and cost efficiency.
When a major APAC bank needed to modernize its manual financial spreading process, Amdocs Cloud Studio delivered an AWS-powered solution that transformed operations, resulting in 90% cost reduction, 52x productivity gains, and 90% accuracy improvement.
Summary
The corporate division of a leading Asia Pacific bank needed to modernize financial spreading – a critical but labor-intensive process for extracting and analyzing financial data from complex documents for credit assessment. With MLOps and GenAI identified as enabling technologies, the bank engaged Amdocs, leveraging Amdocs Cloud Studio to build a solution on AWS. The project resulted in successful deployment of an end-to-end MLOps and GenAI platform that revolutionized financial spreading operations.
- >90% reduction in operational expenditure (OpEx) costs.
- >90% uplift in credit assessment accuracy (especially for net new clients).
- 52x productivity improvement (process time reduced from hours to seconds).
- Governance ensures process compliance and ethical use of GenAI.
Challenge
Labor-intensive process created business risk
Globally, financial spreading is a major pain point for corporate banks. This process involves extracting and transferring data from third-party sources like annual reports into internal financial frameworks. Manual and error-prone, it creates bottlenecks in credit decisions that directly impact revenue and risk management. The process also faces heightened regulatory scrutiny.
The bank’s corporate division needed urgent modernization of this process for credit risk assessment. Specialized accountants typically required over two years of training before they conduct financial spreading analyses. Furthermore, the process for inputting, reviewing, and interpreting complex financial data was time-consuming and demanded high skill levels. Yet outputs weren’t consistent – different accountants often reached different conclusions from identical data sources.
Solution
Advanced cloud tools and platform automates financial spreading
Amdocs developed a comprehensive machine learning operations (MLOps) and GenAI platform using Amdocs Cloud Studio on AWS to automate the bank’s financial spreading process. The initiative took a granular, multiphase approach designed to handle the complexity while ensuring governance and scalability.
The project began by establishing data readiness. This involved developing a data strategy and maintaining ongoing stakeholder engagement to define what the data products would look like. The end goal was confirmed as ‘augmented intelligence’ – where AI and ML outputs support human decision-makers rather than replace them.
Following initial data analysis, the team executed two proof of concept phases: first in a non-production environment, then in production. These implementations generated insights that guided development of the centralized MLOps and GenAI platform now transforming the bank’s financial spreading automation.
Working in a stepwise manner meant the team earned trust and buy-in from the bank’s leadership team and other stakeholders as activity progressed. The approach also reduced risk and meant potential obstacles were identified then resolved on an iterative basis – a major factor in the solution’s successful transition from PoC to deployment.
AWS MLOps and GenAI implementation
The platform combines several AWS tools into a cohesive automated system that demonstrates how financial services providers can implement MLOps at scale. SageMaker provides MLOps capabilities, while Bedrock delivers serverless GenAI frameworks. Meanwhile, S3 handles data storage, and Textract manages document processing. Together, these tools create an integrated MLOps platform with embedded GenAI applications for financial automation.
The end-to-end solution encompasses multiple modules including data compliance and governance, data ingestion, feature engineering, and experiment tracking. These are orchestrated into a unified workflow, with performance and analytics insights available for individual steps and phases, and across the overall business process.
GenAI handles language-based tasks such as document evaluation and text extraction in banking contexts. For example, the model assesses whether new input documents align with established financial document requirements and parameters based on historical data. Suitable documents then proceed to the next stage where relevant content is identified and extracted.
“We have achieved a lot and are now well positioned to solve a hard problem for our business (financial spreading automation) using cutting-edge tech (LLMs), and in the process position us to scale what we have done and how we have done it.”
Banking AI governance and compliance framework
When productionizing data science in banking, the focus typically centers on deployment, automation, and scalability rather than output quality and integrity. However, subjectivity is an industry-wide problem for financial spreading automation with tangible bottom-line impact. To address this, Amdocs introduced strict constraints for data contracts, data quality, and data integrity, supported by a wide range of data engineering pipelines.
For example, some pipelines use AWS Textract to extract data, enrich it via LLMs, then perform quantitative checks to verify accuracy. Each data artifact undergoes lifecycle management during every iteration of the process – ensuring both input and output data meet consistently high standards by applying identical logic and decision-making criteria to each financial spread.
The solution also incorporates strong data and AI governance principles, including working exclusively with datasets that comply with relevant regulatory and banking standards. There’s particular emphasis on embedding explainable and responsible AI (XAI and RAI) to ensure transparency and ethical use. For example, when outputs differ, accountants can now pinpoint precisely when, where, and why discrepancies occur.
The centralized platform seamlessly scales to support new banking and data use cases, which streamlines onboarding new data science applications. Since development costs are amortized over time and use cases, this minimizes short-term profitability impacts while maximizing long-term value. Working in a stepwise manner earned trust and buy-in from leadership while reducing risk – potential obstacles were identified and resolved iteratively, which was a major factor in the successful transition from PoC to deployment.
Outcomes
90% cost reduction and 52x productivity
The MLOps and GenAI platform delivered measurable results across every key performance indicator. Empirical studies comparing the automated solution to manual financial spreading revealed significant improvements in speed, cost, and accuracy that demonstrate how financial services providers can achieve substantial ROI through MLOps implementation.
Manual vs. automated financial processing
Empirical studies on the original manual process revealed that senior accountants could prepare just three to five financial spreads per day, resulting in high costs per spread. There was also an average discrepancy of around 21% when the same spread request was assigned to multiple accountants. For net new clients, this inconsistency increased exponentially.
Using the automated platform, more than 1,300 financial spreads from the bank’s historic, compliance-approved datasets were completed in less than five days. This represents efficiency levels 52x higher than the most senior in-house team members – transforming process time from hours to seconds.
Cost savings and accuracy enhancement
The cost per spread was reduced to less than 10% of manual processing, achieving the targeted >90% operational expenditure reduction. The platform also delivered overall accuracy of more than 90% when working with new, previously unseen datasets. For known clients, accuracy exceeded 91%.
These results represent a greater than 50% reduction in average errors compared to manual processing. The platform eliminated the subjectivity problem entirely – different inputs now produce consistent, objective results regardless of which accountant would have previously handled the case.
Successful MLOps implementation at scale
Although ML model deployment has notoriously high failure rates, Amdocs’ solution achieved successful production deployment. The bank’s corporate division now benefits from a scalable, rapid, and cost-effective solution for financial automation, providing competitive advantage.
AWS experts have recognized the project as innovative, validating both the technical approach and business impact. The platform demonstrates how financial institutions can successfully implement MLOps and GenAI at scale while achieving measurable ROI in banking automation.
