Get in touch

In this edition we continue our 4 Part series, Artificial Intelligence In Action, commissioned from our good friends at propella.ai.

Part 2: Enhancing Payroll Accuracy and Efficiency with Custom Assistants 

This article spotlights a recent case study showing how propella.ai was able to harness AI tools to simplify and accurately complete complicated payroll calculations required by its client On Demand Payroll Solutions.

Client: On Demand Payroll Solutions 

Challenge: On Demand faced challenges in accurately calculating complex redundancy payouts. These types of complicated payments aren’t always handled well by standard payroll systems, putting the accuracy of tax computations at risk. 

Objective: To implement a reliable solution capable of accurately handling complex calculations, particularly for significant redundancy payouts, thereby improving accuracy and efficiency in payroll processes. 

Solution: “Ask Dwayne” which is essentially one of our clients’ safe and secure internal AI Hub. This is a central place that everyone in the organisation can go, to access all the organisation’s AI powered tools. This hub contains custom built assistants (which are hooked into firms’ knowledge base) required by various advisory divisions within the firm (accounting, law, wealth, financial, payroll divisions).  

The custom assistant used by On Demand with their complex tax calculation, is the “Tax Assistant”. The tool is designed to provide employees information about Australian Tax Law from the Australian Tax Office (ATO) and can address and correct any inaccuracies in tax calculations generated by payroll systems. 

Recent Application: 

During a particularly large redundancy payout, the AskDwayne Tax Assistant was utilised to reassess the tax figures initially calculated by the payroll system. The tool provided detailed reasoning and accurate results differing significantly from the payroll software. 

Outcomes: 

• Labour Savings: The reassessment of payroll calculations by the Tax Advisor in AskDwayne significantly reduced the manual effort required to verify and correct payroll inaccuracies, leading to notable labour savings. The reduction in manual errors and increased efficiency in processing complex payrolls also resulted in significant cost savings. Fewer discrepancies mean less time and resources spent on correcting errors and handling disputes, ultimately reducing operational costs. 
• Increased Accuracy: AskDwayne demonstrated superior accuracy compared to other AI solutions, such as OpenAI’s ChatGPT 4.0 mini, by providing correct tax figures and clearly explaining calculation errors from other sources. It provided clear reasoning and source citation, ensuring all parties understand the foundational basis for output. Junior team members can also be trained to use this tool empowering them to handle more complex payroll situations like this with confidence. 
• Efficiency: The tool’s ability to swiftly and accurately handle complex transactions improved the overall efficiency of Demand’s payroll processing, reducing potential delays and errors. 
• Scalability: AskDwayne’s capability to handle complex and large-scale payroll calculations without any performance degradation highlights its scalability. This ensures that as On Demand grows and handles more extensive and intricate payrolls, the tool can scale efficiently to meet increasing demands. 

AskDwayne 6.0 – Tax Assistant has proven to be an invaluable asset for On Demand, particularly in processing complex redundancy payouts. Its superior accuracy, efficient error identification, and clear reasoning set it apart from other AI solutions. The success achieved with this implementation highlights the potential for further improvements and underscores the essential role of AskDwayne in payroll operations. On Demand looks forward to the continuous advancements in AskDwayne, which promises to enhance its capabilities even further. If you are interested in exploring ways to streamline and automate various aspects of business operations, please reach out to Gus McLennan, Managing Director of propella.ai @ gus.mclennan@propella.ai or contact us through our website here.