AI-Assisted Acceptance Testing

Powerlink Queensland is the Government Owned Corporation that owns, develops, operates and maintains Queensland’s high-voltage electricity transmission network: around 15,500 circuit kilometres of lines and more than 150 substations, stretching 1,700 km from north of Cairns to the New South Wales border. Its new Advanced Energy Management System (AEMS), built on OSI’s monarch platform under Powerlink’s Next Generation Network Operations programme, is the platform that will control that network in real time. Before it can do so, it has to pass System Acceptance Testing (SAT).

Challenge

The AEMS programme needed a System Acceptance Testing suite of roughly 3,000 test cases spanning multiple SCADA (supervisory control and data acquisition) domains. Every case had to be technically accurate, fully traceable back to the requirement and design references it verified, and structured so it could be imported straight into Jira Xray, the programme’s test-management tool, without rework.

Authoring at that scale by hand would not fit the timeline. At conventional manual rates of about one quality test case per person per day, three thousand cases represented months of specialist effort that a high-pressure programme did not have. The real constraint was production capacity, and the standard could not be lowered to meet it.

Solution

GIS People built a Microsoft Copilot Studio workflow that accelerates production while keeping technical validation firmly in human hands. Purpose-built Copilot agents are grounded in the requirements, the Monarch AEMS behaviour and the operational context of each SCADA domain, so what they generate reflects how the system actually behaves rather than generic test-writing patterns. Requirement and design references are built into every case as it is produced, so traceability is a property of the output rather than a step added afterwards.

Nothing reaches the suite unreviewed. Each generated batch is validated by a SCADA engineer before it is formatted for Jira Xray import. The workflow runs in four steps:

Outcomes

• A specialised micro team now sustains 70–80 quality test cases a day, around 70 times the output of conventional manual authoring.
• SAT-ready Jira Xray outputs that import directly, with every batch reviewed by a SCADA engineer before it does.
• End-to-end requirements traceability across the suite.
• A reusable, documented prompt library and workflow, ready to extend across further SAT domains.

“This has been a critical piece of work within a very high-pressure program, and the quality, responsiveness and innovation the team has brought to the engagement has made a genuine difference.”

— Powerlink Queensland

Where could AI remove your next bottleneck? We combine AI engineering with domain expertise to automate complex, high-volume workflows where accuracy matters. Talk to GIS People about yours.