By: Mark Coyle
Ask an Australian executive how many people they employ, and you'll get a precise number. Ask how many contractors, labour hire workers and specialist partners deliver their work, and most can't answer. Fewer still can say what those workers are responsible for, how long they've been there, or where the same capability might already exist elsewhere.
For years, that gap in visibility was mainly a workforce and procurement problem. Now AI is making it a work-design problem. For CHROs and other workforce leaders, it’s becoming central to how they make planning decisions, costs, and where AI gets used next.
You can’t automate work you can’t see
Deploying AI well is a work-design problem before it's a technology one. It means breaking a process into its component parts, then deciding which should stay with a person, and which should be augmented or performed by AI. That only works if someone can describe the process as it actually exists today: who touches it, in what order, what outcome they are responsible for, and why.
Most organisations can't, because the workforce delivering that process was never one list in one system. It's permanent staff in the HR platform, contractors sitting in separate supplier arrangements across procurement, and specialist partners engaged directly by individual business units, often with nothing connecting them. The result is an incomplete picture of how the organisation actually operates.
Ask three parts of the same business how a piece of work gets done, and you'll often get three different answers.
That gap already costs money before AI enters the picture: the same capability gets bought three times through three suppliers because no one function can see what the others have, and skills gaps stay hidden until the one contractor who understands the process leaves and takes that knowledge with them.
Workforce planning suffers for the same reason: it’s hard to decide where to redeploy people or invest in new skills when no one can say, with confidence, who’s doing what.
AI adoption raises the stakes, because automation depends on understanding work at a much more granular level. MIT researchers studying enterprise AI adoption found 95 per cent of generative AI pilots deliver no measurable financial return, attributing much of the problem to a "learning gap", where businesses are layering AI onto processes they had never properly mapped or understood.
Agents are already joining the workforce
It gets more complicated from here. AI agents are becoming a category of labour in their own right. The federal government's AI policy now requires agencies to assign an accountable officer to each AI system and log it in a register, introducing a level of governance similar to that applied to a person doing the work.
Once an AI agent takes on tasks previously handled by a person, organisations need to decide who is accountable for its output and how that work is governed. Businesses that can't produce a clear picture of the humans doing their work are now being asked to make that same set of decisions for AI agents, on a foundation that was never laid out.
That's the real cost of the visibility gap. No one can make a deliberate call on the mix of specialist partners and AI agents doing the work, because no one has an accurate view of the mix already in place. Every decision about what to automate and what to keep human is being made on partial information.
Regulators have already worked this out
The government isn’t waiting for businesses to catch up. A draft review of the Closing Loopholes Acts, released in draft in May, found that more than 104 Same Job, Same Pay orders have already delivered wage increases of up to $60,000 a year for over 8,000 workers, while reinforcing that worker classification depends on the day-to-day reality of a relationship.
With award wages up 4.75 per cent and the minimum wage up 6 per cent from July, a misclassified worker is a more expensive mistake to make than it was twelve months ago. And with Payday Super putting super payments on the same clock as wages, reported to the ATO within seven business days, the regulator now has a near real-time view of who's being paid, and how. Increasingly, it will find the gap in a workforce before its employer does.
Treat visibility like infrastructure
None of this means stepping back from contractors, specialist suppliers, or AI. All three are becoming more central to how work gets done. But the organisations that navigate that shift well will be the ones that treat workforce visibility as core infrastructure: maintaining a current view of who is doing the work, whether they're correctly classified, what they cost and where an AI agent could realistically take on part of the workload.
In a market racing toward AI adoption, that visibility is no longer optional. Get it right, and the decisions that follow, on compliance, workforce planning, cost and automation, get easier. Get it wrong, and the regulator will find the gap in your workforce before you do.
About the author: Mark Coyle works as the CEO- APAC at HeadFirst Global. He focuses on helping organisations navigate evolving workforce needs through technology, talent and flexible workforce models. He is passionate about building high-performance cultures that deliver customer-centric solutions. HeadFirst Global brings together specialist talent, AI, data and technology through a global work orchestration platform designed to help enterprises improve workforce outcomes.