Leading Professional Services Firm Scales Financial Analysis with AI

About the Client

The client is a global professional services organization, serving companies across a wide range of industries in markets around the world. Its teams support complex client engagements where financial analysis is central to the value they deliver.

Goals

The firm’s client engagements rely on critical data contained in financial documents. Key fields are spread across both structured tables and unstructured content such as footnotes and attached contracts, which each company formats differently. Extracting them is a time-intensive, detail-heavy process with little margin for error, requiring significant time from skilled analysts whose expertise is best spent analyzing what the data means.

The main goal was to extract these fields with speed and precision, so analysts could focus on high-value work. To get there, they aimed to:

  • Train and deploy AI models able to extract the right information from all the different types of documents
  • Classify each document by type with high confidence, including new formats never seen before
  • Achieve high precision in the extracted fields
  • Give the business team an efficient way to validate the extracted data

Solution

Together with Bridgenext, the client’s team used Generative AI to build an advanced solution for extracting and analyzing information from financial reports. The work progressed in phases.

  1. Classification

    Generative AI produced a classification dataset, in which a series of documents were labeled according to their specific types. This dataset then trained a classification model that categorizes new, unseen documents with a high confidence level, so each report enters the right extraction process.

  2. Extraction

    The joint team developed prompts specifically designed to extract the correct fields from the already classified documents. In an iterative process, both the prompts and the model were continuously tested and improved, with the goal of increasing precision in the outputs.

  3. User Acceptance Testing

    The solution went through iterative User Acceptance Testing with its final users: the business team. Their deep business knowledge helped determine whether the precision of the extraction was high enough or if it could be improved. The business team now works in a user interface embedded in Microsoft Teams, where they can review and edit the extracted fields, or even adjust the prompts to refine the extraction process. Once the required data is ready, it can be exported to a CSV file and made available for analysis and sharing as needed.

Results

The AI-driven solution marked a significant milestone for the client, addressing a challenge that had persisted for nearly a decade and demonstrating the scalability and versatility of the firm’s AI capabilities.

  • High-Volume Automation: The solution processes between 3,000 and 7,000 documents annually, automating the extraction of critical information from financial reports.
  • Greater Accuracy and Reliability: AI streamlined the process while enhancing the accuracy and reliability of the extracted data.
  • Business Team in Control: Users validate and refine the extracted data directly within Microsoft Teams, the tool they already use every day.
  • Expanding Capabilities: Building on this success, the client is now extending the solution to include an agreement analyzer, projected to save approximately 3,100 hours annually.
  • Expertise Where It Matters Most: With extraction automated, analysts now dedicate their time to high-value analysis, expanding the firm’s capacity to serve client engagements.

Industry

BFSI – Professional Services

Benefits

  • 3,000-7,000 documents processed per year
  • Higher accuracy and reliability in extracted data
  • Expert time now focused on high-value analysis
  • ~3.1k hours a year in projected savings as the solution expands