The challenge of detecting errors in complex statements

December 28, 2020

Bob Suh, CEO OnCorps

The Burden of Draft Reviews

The burden of checking complex and lengthy financial and regulatory reports is handled in various ways by asset managers. Some choose to have internal staff print and check prepared reports. Others pay consultancies and business process outsourcers to manually error-check documents. Still others rely on the report preparer to check accuracy with little independent oversight.

A Comprehensive AI Solution

The core platform can be configured to run checks specific to each report (see illustration below). It is configured to know which sections to check, what specific algorithms to apply, and whether to cross-validate the data with other sources either in the report or in other data sources. As breaks are found in the checks, the system will classify the causes of the breaks to help train the algorithm and improve predictions.

The OnCorps team was asked to build an AI platform that would reduce as many hours as possible by error checking a complex and lengthy semi-annual financial statement. The report was over 400 pages in length and needed approximately 106,000 checks by the algorithm. We achieved a dramatic reduction in labor hours. The oversight team's prior period hours were substantial and required reviews of three drafts. When we completed coding all the checks, we had no idea how long it would take to review the entire document. When we hit the return key, we were amazed to see it complete all 106,000 checks in under 40 minutes.

Featured Content

August 12, 2021
Predicting NAV delays with AI
Read More
July 27, 2021
Decision Simulations
Read More
January 11, 2021
Data Challenges in Adopting AI
Read More

Interested in discussing a solution?

Get in touch