Position Descriptors Assistant streamlines staff classification
A manual process that once consumed hours of recruitment team time has been transformed through AI-powered automation. UNE's Position Descriptors Assistant demonstrates how targeted AI solutions can address specific operational challenges, while maintaining the precision required for critical processes.
The assistant emerged from a clear operational need identified by UNE Senior Manager Talent Acquisition Sarah Stuhmcke. "The professional staff position classification process is critical to ensure that professional staff positions are appropriately classified and therefore, the correct pay rate is applied," Sarah explains. "We didn't have any tools or systems to assist with this. It was a time-consuming, manual process for the recruitment team."
The stakes are significant. Incorrect classification means UNE staff may not be paid correctly for the nature of work they perform, making accuracy essential.
Building precision into automation
The assistant follows a rigorous three-step methodology designed to replicate and enhance human classification expertise. First, it extracts criteria exclusively from two specific sections of position descriptions: "The Role Key Responsibilities" and "Selection Criteria."
Next, the assistant classifies each criterion individually against UNE's Enhanced Descriptor Tables for Levels 1-10. Each classification includes explicit justification, directly quoting relevant phrases from both the position description and the corresponding descriptor table. When ambiguity exists, the assistant identifies secondary classification levels and explains its reasoning. Finally, it synthesises individual criterion classifications to provide an overall recommended classification level for the entire position, highlighting any significant variations that influenced the recommendation.
Iterative development and testing
After an initial build in Cogniti, the assistant was migrated to Madgwick to allow it access to a wider scope of models. "Once the tool was built in Madgwick, it gave us a fantastic starting point to test and evolve the tool with LabNext70," Sarah says. "Shannon (Tyrrell) from LabNext70 led the technical development, working closely with the entire recruitment team. "All recruitment team members were involved in our testing plan and thoroughly tested each version of the assistant to ensure it was fit for purpose and accurate."
Measurable efficiency gains
The results demonstrate clear operational benefits. "Each position classification would take up to 2 hours of recruitment team time," Sarah reports. "Utilising the Position Description Assistant has halved this time."
This 50% time reduction allows recruitment staff to focus on higher-value activities while maintaining the accuracy essential for proper staff compensation. The assistant serves as a practical example of how AI can enhance human expertise in complex administrative processes.