Initiative overview
The DRA was developed to increase efficiency in access to records within a specific business area of the Department of Finance (Finance). Previously, staff relied on manual processes to review documents and record key details, which created barriers to consistent record keeping and made it difficult to efficiently research existing information. DRA was introduced to improve access to historical records and reduce administrative effort.
The initiative combines AI with automated workflows to support staff with clearly defined tasks. When a document is saved to a designated system, DRA extracts relevant information and presents it to users for review through a structured approval process. Staff remain responsible for reviewing and approving the information before it is recorded, ensuring appropriate oversight and accuracy at all times.
The objective of DRA is to improve consistency in how documents are recorded, reduce duplication of effort, and support better use of existing information within the business area it serves. By capturing information in a structured and standardised way, the initiative helps staff more easily conduct research and apply existing resources to current work. The system has been assessed as low risk, as it does not interact with the public, make decisions independently or process personal information.
DRA is deliberately designed as a targeted solution tailored to the needs of a specific business area, rather than an enterprise-wide system. It reflects the Finance’s approach to using AI for well scoped, low risk, assistive purposes where there is a clear operational need. It represents the type of AI enabled solution Finance currently uses and plans to expand further in 2026 by applying similar approaches to other suitable internal processes.
Results, outcomes and impacts
The initiative has delivered measurable productivity improvements while maintaining human oversight. Results were assessed through A/B testing, which showed that using DRA reduced the time required to extract and record key information by approximately 15–30 minutes per document compared to manual methods. When applied across typical annual workloads in the business area, this represents a meaningful reduction in administrative effort each year. Over time, these improvements are expected to strengthen record keeping, improve access to historical records, and support more consistent use of existing information resources.
