SHREVEPORT – Accountants all across the country are uncrossing their eyes and unscrambling their brains with the United States tax filing deadline coming and going.
Poring over bank statements, analyzing spreadsheets and reconciling clients’ accounts requires an eagle eye for detail along with tons of manual data entry and transaction-by-transaction checking.
LSU Shreveport computer science professor Keyvan Shahrdar has developed a tool to lighten the load on accountants.
Meet ChatBotCPA, a computer program that harnesses artificial intelligence and advanced natural language processing and machine learning to understand – and complete – tasks.
ChatBotCPA can compare multiple documents like bank and credit card statements to a company’s QuickBooks database, highlighting missing entries or discrepancies.
An accountant can ask this intelligent assistant to then reconcile these accounts – anything from providing a list of discrepancies to actually filling in the missing information on the spreadsheet.
“Accountants can delegate repetitive and time-consuming tasks, such as data entry, expense categorization and transaction matching,” said Shahrdar, who founded Shahrdar Enterprises LLC to pitch ChatBotCPA and other products to potential clients. “This frees up time for accountants to focus on more complex tasks and provide better service to their clients.
“ChatBotCPA can streamline all kinds of processes and also eliminates the need for manual data entry, which significantly reduces the effort involved and improves data accuracy.”
Shahrdar generated the idea when his wife, accountant Laura Shahrdar, was describing the time-consuming steps in current accounting practices.
“Accountants aren’t computer scientists – they aren’t familiar with the intricacies of handling and sorting big data,” Shahrdar said. “So I started toying with this idea when I noticed the inefficiencies and manual work involved in the accounting process.
“My wife may spend a couple weeks processing a client’s account, whether that’s waiting on the client to send certain statements or manually processing the data. ChatBotCPA can do that in a couple of minutes.”
Laura Shahrdar’s tax season operated more smoothly than in years past as she delegated many of her repetitive tasks to the chatbot.
“This conversational processing automation bot handles various tasks that would otherwise consume a significant amount of time, such as reconciling accounts and tracking transactions,” said Laura Shahrdar, CEO of Shahrdar Enterprises. “This allows me to focus on more complex and high-level financial analysis.
“It has helped me manage a larger client base by automating these tasks. Many of these tasks are prone to human error because they are repetitive and complex … but the chatbot can accurately and consistently perform these tasks because of its built-in algorithms and machine-learning capabilities.”
Laura Shahrdar said other typical functions in which the ChatBotCPA assists is flagging discrepancies between different data sets, tracking and categorizing expenses, generating financial reports and assisting with tax calculations and projections.
The ChatBotCPA’s development process started six months ago, beginning with the initial coding and programming of the system over a period of three weeks.
Keyvan Shahrdar continued his tinkering by incorporating feedback from users and repeatedly updating ChatBotCPA’s capabilities.
Phillip Kilgore, a former LSUS professor who is a machine learning expert, said ChatBotCPA uses a natural language interface to understand and complete a command an accountant might give.
Kilgore uses Star Trek captain Jean-Luc Picard and the USS Enterprise to explain how the accountant and ChatBotCPA work.
“Captain Picard doesn’t program the Enterprise’s computer, he just says what he wants, and the computer susses out the intent and makes it happen,” said Kilgore, who serves as Shahrdar Enterprises’ chief technology officer. “The accountant is like a task master.
“Accountants aren’t spending as much time figuring out how to achieve a task. Instead, they are focusing on actually realizing the result -- that is determining the materiality of what ChatBotCPA did. Accountants can make more informed decisions because they are spending less time doing the nitty gritty.”
Kilgore said because machine learning has been around for decades, programs like ChatBotCPA are accurate because they complete tasks with rote algorithms – meaning the process doesn’t change.
Accountants can discern meaning and importance from data provided by ChatBotCPA while also staying abreast of the constant legal changes.
The ChatBotCPA is also capable of communicating with clients, sending regular updates about the client’s account status via email.
“The bot’s ability to interact with clients has streamlined communication, reducing the time spent on back-and-forth emails and phone calls,” Laura Shahrdar said. “It allows accountants to focus on more critical aspects of their job, ensuring a higher level of overall service quality.”
Keyvan Shahrdar said he’s received interest from a top 25 accounting firm in the U.S. among other potential clients.
