UNE's AI financial assistants streamline transaction coding

Staff across UNE now have access to two powerful AI assistants that take the guesswork out of financial coding. Developed by Sharon Smith, Purchasing Officer in Financial Services, Business Intelligence (FSBI), these tools help staff quickly find the right codes for financial transactions - from requisitions and purchase orders to credit card acquittals.

Two complementary assistants for a complete solution
The financial coding process at UNE requires two distinct pieces of information: who is spending the money and what they're spending it on. To address this, Sharon's team has created two specialised assistants:

Chart of Accounts Wizard identifies the correct departmental codes - answering "who is spending the money." It provides the Division, Directorate, and Cost Centre codes specific to your area.

Natural Accounts Wizard determines the appropriate expense codes - answering "what you're spending money on." This assistant helps classify purchases like stationery, equipment, or services.

"Not everyone knows where to look to find those codes or knows which code they should be using, so we've tried to make that a lot simpler," explains Sharon.

Simplifying complex financial processes
The Natural Accounts Wizard tackles the more complex side of financial coding. When staff are uncertain which expense code to use, they can simply describe their purchase.

"You could put in 'code for stationery,' and it'll have a think and come up with all the different matches for that you might use for a stationery order," Sharon demonstrates.

The system even handles edge cases. During a recent office discussion about how to code a first aid kit purchase - an item not explicitly defined in the source documents - the team enhanced the assistant to provide the correct guidance.

Building solutions from staff experience
The creation of these assistants reflects Sharon's practical approach to problem-solving. Having run UNE's Purchasing Training for several years, she consistently heard the same feedback: understanding the Chart of Accounts was the most challenging aspect for staff.

"The content about understanding UNE's Chart of Accounts and how it fits in with other processes is the most useful," Sharon explains.

Sharon worked with the LabNext70 team to create the assistants, and found the solution didn't require complex AI engineering. Both use simple prompts - less than 100 words each - combined with Knowledge documents containing the relevant financial data. The Chart of Accounts Wizard draws from UNE's official Chart of Accounts PDF, while the Natural Accounts Wizard interrogates the Natural Accounts spreadsheet.

"UNE has always had a pdf of the Chart of Accounts, and a spreadsheet of the Natural Accounts available on the web, however not everyone knows where to find them," Sharon notes. "The Natural Accounts document is quite large, and about half of it doesn't apply to staff outside of finance so it can be confusing trying to choose the correct code."

The assistants represent what Sharon calls a "see a need, fill a need" solution. Rather than requiring staff to navigate complex documents, the AI handles the cognitive load of understanding and searching through the information, presenting only the relevant results.

Accessible and user-friendly
Both assistants are designed with simplicity in mind. Users don't need to formulate perfect questions or use specific terminology. As Sharon notes, "You don't even have to be polite. You can quite literally just type in the text of what it is that you're looking for."

For the Chart of Accounts Wizard, the system provides comprehensive information including Division, Directorate, cost centre, and whether the code is active.

The success is evident in the usage data. "Every time I log into Madgwick I can see they are both in regular use and I haven't had any complaints," Sharon observes. "I have to assume they are doing their job well." 

Where to access the assistants
Staff can access both assistants through the UNE financial services website or directly through the Madgwick platform.