Fix the constraint, not the workflow.

Every delivery team has one queue that sets the pace of everything behind it. In most RTOs, that queue is the marking pile.

Results wait on marking, resubmissions wait on results, completions wait on resubmissions, and enrolment planning waits on all of it. You don’t fix that by redesigning the chain. You find the link that governs it and widen that.

Ask where work waits longest and the answer in vocational education is rarely the classroom. It’s the pile on the assessor’s desk. High volume, repetitive reading, and one assessor allowed to clear it. Whatever that queue can process per week is what your whole RTO can process per week, which makes it the single highest-leverage place to put better tooling.

Inside the trainer’s estimated 16.25 minutes per assessment by hand, most of the time isn’t judgement. It’s reading, cross-referencing the guide, hunting evidence through templates, and writing feedback from a blank box. Those are transferable hours. A machine can do the laying-out. The minutes that remain, the borderline calls and the final say, are precisely the ones that should stay with a person. For one trainer, splitting the work that way took an assessment from her own estimate of 16.25 minutes to 5, on a batch she timed herself.

The pile is a queue, and queues obey maths, not effort.

The test of a workflow change isn’t the demo. It’s the queue. Run one unit, one trainer, one term and read the system’s own numbers: days from submission to result, assessor-hours per batch, consistency from the first assessment to the last. If the queue didn’t shrink, the tool didn’t work, and no feature list should talk you out of that reading.