AudiTech

Inspiration

Being from a Chartered Accountant background, one of our teammates analyzed manual processes involved in the auditing process in income tax validation. We performed Root cause analysis on this issue and found out that depending on the organization turnover there is more and more utilization of resources(Article Interns) for examination or inspection of various books of accounts followed by physical checking of inventory to make sure that all departments are following the documented system of recording transactions. It is done to make sure the accuracy of financial statements provided by the organization to avoid any fraudulent scams while submitting tax. The process is tiresome and monotonous since Auditors must go through each and every receipt generated by the organization within one financial year. Hence, we decided to implement innovation in this particular domain to avoid fraudulent transactions by automating the process from start to end.

What it does?

The process of going through every receipt is prone to manual error. Thus, it has become imperative to automate specific tasks in auditing. As a part of our automation, we have developed an app which reads the details from the set of bills and segregate and extracts the relevant data which is required to generate the “General Ledger” which contains useful information like Total Amount of money credited or debited in the transaction, Date of the transaction and the Name of the organization. The Owner of the organization submits a General Ledger to Chartered Accountants, and the generated General Ledger by Auditors is then verified for any missing transaction records to identify any fraudulent transaction in the organization.

How we built it?

Our journey started with analyzing all the manual processes involved in this process of auditing by consulting with one of the Chartered Accountants and understanding all the challenges they are facing in this validation. After that, we started working on the conversion of Image to Text, and we decided to go ahead with Microsoft Azure’s Image to Text conversion API and integrated this with an Android application to make the user interface more efficient and more comfortable to use. Integrated all the processes since Azure’s API is developed using Python, and an android application was developed using Core Java.

How does it work?

Instead of multiple resources, we developed an android application where a single resource can click pictures of all the receipts generated by the organization within that financial year. Once the user clicks all the images, he needs to trigger the process from the application, and our tool will automatically read all the necessary information from the receipts and generates corresponding General Ledger and will also validate the generated General Ledger with the one submitted by the owner of the organization. Once the validation is done, the application will tell the user if there is any discrepancy in the financials generated by the organization with the corresponding record number, thereby making trace back more feasible.

Challenges faced?

Our one of the biggest challenge was achieving the maximum accuracy of the Image to Text recognition because of blurry images. To overcome this issue, we pre-processed the images before utilizing Azure’s API to achieve better efficiency.