Project Overview
A governmental agency experienced a delivery backlog because of large volumes of complicated documents that needed to be processed.
By integrating PaperEntry AI into its data entry operation, Deep Cognition enabled approximately 75% time savings on evidence gathering with 99%+ data accuracy. This boosted the agency’s case capacity and significantly reduced their document processing backlogs.
Results
Time Savings
Using PaperEntry AI, Deep Cognition completed this project in only six weeks, providing a time savings of approximately six months over the previous process with the internal tool.
Cost Savings
Deep Cognition’s solution provided additional savings by eliminating the costs of having governmental agency personnel run the internal tool and validate and correct the resulting data.
Ease of Use
Providing the digital output in a simple Excel format allowed the agency to efficiently expedite the large volume of reconciled and validated data.
99%+ Data Accuracy
PaperEntry AI’s high accuracy combined with turn-key human validation achieved near-perfect results (>99% accuracy).
The Problem
Shortage of Workforce
Manual Data Entry
of Unstructured Data
Limited Processing Power
of Internal Software Tool
A governmental agency has a complicated and voluminous case to investigate a business for fraud. The document workload contained more than 100k documents from the company’s bank. The files contained checks, money orders, correction slips and deposit tickets.
The data from these files needed to be extracted and stored in a usable digital format to thoroughly analyze and investigate the company’s financial activity. Each deposit ticket also needed to be reconciled against the corresponding check totals to ensure all the data was captured accurately and properly accounted for.
This was the third round of this type of case. For the first two rounds, the agency attempted to use an internal software tool to complete the projects. Some of the challenges it faced included:
The governmental agency did not have the human resources available to manually extract and store the large volume of data accurately and in a timely manner.
Many checks were handwritten and difficult to read and the data was unstructured.
The internal software tool took more than 7 months to prepare its final results because of its limited capabilities, such as:
- Inflexible template based approach
- Limited processing power and memory
- Time consuming and difficult validation
- High amounts of manual corrections
Summary of the main project requirements for the governmental agency:
- Quick and accurate data extraction and validation
- Scalability
- Little human intervention in processing larger volumes of data
The Solution
For the third round, Deep Cognition was selected to complete the project, which resulted in a much faster and more efficient process.
The government agency sent Deep Cognition the electronic files it had received from the company’s bank. Deep Cognition completed the project in record time using PaperEntry AI.
The Implementation
While the project was underway, the Deep Cognition validation team noticed the files contained thousands of money orders and some correction slips in different formats from the various checks.
Since PaperEntry AI is not template based, Deep Cognition was able to handle all the new formats and accurately process the money orders and correction slips.
- PaperEntry AI’s built-in, state-of-the-art OCR and proprietary AI engines quickly and accurately extracted and validated the document data, including check number, payee, payor, date, amount and memo.
- PaperEntry AI reconciled the deposit tickets to ensure the dollar amount on each ticket equaled the total amount from its corresponding checks.
- PaperEntry AI provided the agency additional time and effort savings by:
- Normalizing payee and payor names (i.e., grouping all the variations of the same name together)
- Providing hyperlinks back to the original documents
- Providing advanced ways to sort and filter the results
- Deep Cognition delivered final results that met the high level of accuracy required by the government agency.
- Deep Cognition provided weekly updates to the governmental team regarding: progress, completion dates, challenges, and resolutions.