In today’s competitive banking environment, effective credit assessment is essential to ensure sound lending decisions and minimize default risk. As businesses grow more complex, financial institutions face increasing challenges in evaluating loan applications — from reconciling multiple documents like company profiles, financial statements, and credit reports, to interpreting key ratios that indicate a company’s financial health.

Traditionally, this process is manual, time-consuming, and prone to inconsistencies. Credit officers often spend hours reviewing documents, re-entering figures into spreadsheets, and calculating ratios such as current ratio, debt-to-equity, and return on equity. These inefficiencies not only delay loan approvals but can also lead to errors in judgment — either rejecting viable borrowers or approving loans that carry hidden risks.

Regulators and central banks, including Bank Negara Malaysia, place strong emphasis on prudent credit management, requiring financial institutions to demonstrate robust credit policies and reliable evaluation methods. This makes accurate and timely financial spreading — the process of extracting, standardizing, and analyzing financial data — a cornerstone of responsible lending practices.

However, the manual approach to financial spreading is no longer sustainable. With increasing volumes of loan applications, banks need solutions that can process structured and unstructured data quickly, reliably, and at scale.

To address this, organizations are adopting Agentic Automation — combining UiPath Autopilot, Document Understanding, and intelligent workflows to replicate the work of a credit officer. In this setup, officers only need to upload the required documents, while Autopilot takes over the complex steps: validating shareholder identities, extracting company information, analyzing multi-year financials, cross-checking credit reports, and generating loan recommendations.

This demo showcases how financial spreading for loan applications can be streamlined end-to-end with assisted automation — reducing manual effort, ensuring accuracy, and producing structured outputs in Excel format for faster and more consistent decision-making by the credit committee.

Financial spreading is a critical process in credit risk management, ensuring that lending decisions are based on accurate, consistent, and well-analyzed data. Banks and financial institutions rely on spreading to transform raw financial statements, company profiles, and credit reports into standardized formats for comparison and decision-making.

However, in many organizations this process remains manual and fragmented. Credit officers must collect multiple documents — such as company SSM profiles, shareholder identity cards, five-year financial comparisons, and CTOS/CCRIS reports — often in varying formats like PDFs or scanned images. Each document must then be reviewed, with key details such as shareholding structures, revenue growth, liabilities, and arrears extracted and keyed into spreadsheets. Officers must also calculate financial ratios, check for arrears, validate ICs against company profiles, and reconcile discrepancies across sources.

While necessary for sound lending, this people-intensive approach creates bottlenecks. A single loan application can take hours or even days to review, and human error is a constant risk — from mis-entered figures to missed red flags such as unsettled arrears or litigation history. The downstream impact includes delayed loan approvals, inconsistent recommendations, potential credit losses, and higher operating costs as banks increase headcount to keep pace with loan volumes.

This was precisely the challenge faced by the credit team in this case. Each application triggered a long cycle of document handling, validation, and manual analysis, making it increasingly difficult to ensure timely and consistent loan approvals at scale.

What the organization required was a smarter approach — one that could leverage UiPath Autopilot to handle document intake, extraction, and validation automatically, surface missing or mismatched information in natural language, and provide clear financial analysis and recommendations. Once validated, users could trigger an attended automation to generate a structured Excel report, ensuring an end-to-end workflow that combines speed, accuracy, and reliability in credit assessment.

The primary objective for the financial institution was to transform its credit assessment process into a more intelligent, reliable, and scalable workflow. As loan application volumes increase and regulatory oversight strengthens, the institution recognized that traditional, manual financial spreading methods were no longer sustainable. They sought to digitize and automate the end-to-end loan assessment process — from document intake and validation to ratio analysis and loan recommendation — using a combination of UiPath Autopilot, Robotic Process Automation (RPA), and intelligent workflows.

The institution’s vision was to reduce dependency on manual data entry and spreadsheet calculations, enhance the consistency and speed of loan evaluations, and establish a framework that could adapt dynamically to new credit policies, risk parameters, or regulatory requirements.

Key objectives and goals included:

• Automate financial spreading: Use Autopilot to handle document intake, extraction, and cross-validation of ICs Automate financial spreading: against SSM company profiles.
• Improve decision accuracy: Leverage AI-driven data analysis to calculate financial ratios, identify arrears, and highlight litigation risks with greater consistency.
• Streamline credit checks: Integrate CTOS/CCRIS data with internal financial comparison documents to ensure complete and timely assessments.
• Reduce manual workload: Eliminate repetitive tasks such as data entry and ratio calculations, allowing credit officers to focus on judgment and approvals.
• Ensure regulatory compliance: Maintain audit trails of extracted data, validations, and recommendations to align with Bank Negara Malaysia’s credit risk management guidelines.
• Enable scalability: Deploy a modular, low-code automation framework that can support higher loan volumes, new product types, or updated credit risk models without major rework.

To tackle the financial institution’s credit assessment challenges, Cybiant designed an intelligent Financial Spreading solution leveraging the UiPath RPA platform combined with advanced AI and Autopilot capabilities. Central to the solution is an intelligent agent powered by a Large Language Model (LLM), contextualized on common credit assessment practices used by Malaysian banks, and aligned with Bank Negara Malaysia’s credit risk management expectations. This ensured that the agent’s analysis and recommendations were automated while still reflecting local lending standards and prudential requirements.

Cybiant developed a scalable intake process capable of handling multiple document formats — including SSM company profiles, shareholder ICs, five-year financial comparison statements, and CTOS/CCRIS credit reports. This pipeline enables Autopilot to automatically extract, validate, and normalize financial and identity data, reducing dependency on manual review. By embedding the intelligent agent within UiPath RPA workflows, Cybiant enabled an end-to-end assisted automation — from document collection through to financial analysis and structured output in Excel.

To complement this, Cybiant configured Autopilot to perform consistency checks such as verifying IC numbers against the SSM profile, detecting missing shareholder documents, and validating shareholding information. These validations are handled intelligently by the agent, ensuring data integrity before moving on to financial analysis. UiPath Studio was then used to develop the automation that converts Autopilot’s structured output into a professional Excel file, ready for review by the credit committee.

Key steps Cybiant took to create this solution included:

1. Credit Risk Criteria Definition and Contextualization: Defined key lending parameters, including liquidity and leverage thresholds, arrears tolerances, and internal credit risk indicators. These were embedded into the agent’s prompt so that recommendations aligned with both internal policies and Bank Negara Malaysia’s credit risk management expectations.
2. Agent Development: Built an intelligent agent on the UiPath Autopilot platform that leverages the contextualized LLM to analyze financial ratios, arrears, and credit scores, and to provide loan recommendations with transparent reasoning.
3. Data Ingestion Pipeline: Engineered a robust ingestion system capable of processing SSM, IC, and credit report documents in various formats (PDFs, scanned images, structured data), allowing Autopilot to normalize inputs for accurate analysis.
4. Consistency Checks via Autopilot: Configured Autopilot to validate IC numbers against company records, detect missing or incomplete shareholder documents, and ensure shareholding details match the SSM profile.
5. Excel Automation via UiPath Studio: Built a workflow in UiPath Studio that takes Autopilot’s structured output and generates a professional Excel file with three tabs: Key Information, Core Financials, and Financial Analysis & Recommendation.
6. Testing and Validation: Conducted extensive testing on historical loan applications to validate ratio calculations, ensure data extraction accuracy, and confirm alignment with the bank’s lending policies.
7. Deployment and Monitoring: Successfully deployed the solution into a demo environment, with ongoing monitoring and adaptive improvements to accommodate evolving credit policies, inflationary factors, and business trends.

Cybiant deployed a strategic combination of Robotic Process Automation (RPA) and intelligent agent technology to automate and enhance the financial spreading process for loan applications. The core automation was developed using UiPath, which orchestrates the end-to-end workflow, including document intake, CSV-to-Excel conversion, and structured output generation for credit committee review.

At the intelligence layer, Cybiant integrated an agent powered by a Large Language Model (LLM) through UiPath Autopilot. The model was guided by a carefully engineered prompt designed for the credit assessment context of Malaysian SMEs. This prompt reflects typical lending parameters used by banks and the credit risk management expectations set by Bank Negara Malaysia. It instructs the agent to validate IC numbers against the SSM profile, detect missing shareholder documents, extract multi-year financial data, and calculate ratios such as current ratio, debt-to-equity, and return on equity.

The agent evaluates uploaded documents including company profiles, identity cards, financial comparison statements, and credit reports (CTOS/CCRIS). It cross-references key data points — for example, matching shareholder ICs to company records — and generates concise outputs such as:

• “IC validation successful — 2 shareholders matched with SSM profile.”
• “Financial ratios calculated: current ratio 5.0, debt-to-equity 0.19.”
• “Arrear detected: 1 month on secured car loan facility.”

These AI-powered validations and analyses are embedded within the broader UiPath workflow, enabling a smart, efficient, and auditable credit assessment process. This approach significantly reduces manual effort while improving reliability and transparency. Together, the UiPath automation and Autopilot agent form a cohesive financial spreading automation system that aligns with local lending standards and supports consistent, data-driven loan decisions.

Cybiant’s Agentic Automation solution transformed the institution’s credit assessment and financial spreading process. By combining UiPath Autopilot, prompt-based financial analysis, and RPA workflows, the institution achieved faster loan evaluations, higher accuracy, and greater scalability – without the overhead of manual data entry or spreadsheet-based analysis.

Key results and benefits included:

1. Streamlined Financial Spreading: Automated the extraction and structuring of SSM profiles, shareholder ICs, financial statements, and credit reports into consistent, analyzable formats.
2. Faster Loan Assessment: Reduced processing time for each loan application by automating document validation, financial ratio calculation, and arrears checks.
3. Improved Accuracy and Consistency: Leveraged prompt-engineered AI to apply the same credit assessment rules across all applications, reducing subjectivity and human error.
4. Integrated Credit Analysis: Combined CTOS/CCRIS credit scores with multi-year financial data to provide a holistic and transparent loan recommendation.
5. Configurable Credit Policies: Enabled credit teams to adjust thresholds (e.g., liquidity ratio, debt-to-equity limits) directly in the workflow, aligning loan recommendations with evolving risk appetites.
6. Reduced Operational Load: Minimized manual spreadsheet work and document reviews, allowing officers to focus on higher-value judgment and decision-making.

Overall, Cybiant’s solution empowered the institution to run smarter, faster, and more reliable credit assessments. The use of practical AI — guided by real-world lending policies rather than complex model retraining — ensured a scalable, audit-ready solution aligned with Malaysia’s credit risk management expectations.