Australian employees are not waiting for the AI strategy. 78% of Australian AI users are already bringing their own tools into work without approval, and 84% of Australian workers are using AI at work, higher than the global average and rising (Microsoft & LinkedIn, 2024). The AI work redesign is happening on weekday afternoons, in tools paid for personally and used informally. There is no roadmap. The structural decisions this produces sit above HR and IT — they are the executive’s call, not an advisory question.
In Brief
- The AI work redesign is already happening in most Australian organisations — eight in ten employees are bringing their own tools into work without sanction.
- The people closest to existing processes are making structural choices about which work AI will absorb — choices that lock in commitments only the executive can change later, at higher cost.
- The default executive response of educating the workforce into existing structures is being made twice as often as the response of redesigning the structures themselves (Deloitte Australia, 2026).
- The structural decisions sit above HR and IT — sanctioning, decision rights, capital allocation between training and redesign, and the call on what the organisation will stop doing.
- The executive who makes those calls early sees a different organisation eighteen months later than the one who delegates them.
This is a structural condition, not a governance lapse. When the tools are this accessible and the productivity gains this immediate, the people closest to the work redesign their own work first. They reorganise tasks, retire steps, change how they collaborate, and decide what the next version of their role looks like, all before the executive has approved a paper on the question. By the time the executive turns to AI in the operating model, the shape of that operating model has already started to shift, in directions that were not chosen at the executive level.
The redesign is already underway
The condition the executive is now operating in is structural, not cultural. Across organisations doing many things right, AI adoption arrives through individuals before it arrives through any plan. Each employee makes a small set of decisions about what to delegate to a tool, what to keep, and what to stop doing. None of those decisions is wrong on its own. Cumulatively, and over time, they amount to a redesign of the area’s operating model. The work that used to flow between three roles now flows through one role and a tool. The handoffs that used to coordinate two teams are being absorbed into a single individual’s prompt. The decisions a manager used to make about scope and quality are now being made by a system the deputy did not choose and cannot inspect.
The people doing this redesign are doing it well, within the limits of what they can see. What they cannot see is the structural layer. Legal exposure is created when customer data passes through a tool whose terms of service the organisation has not read. The audit trail will be missing when the board next asks who decided. A capability gap is being built quietly into the next generation of leaders, who are skipping the experiences that used to produce judgement because AI is doing the work the experiences used to teach. And eighteen months from now, when a competitor restructures around a redesigned operating model, the organisation that left the question alone will find itself locked into the accumulated micro-decisions of three years of unsanctioned adoption.
The reason the redesign is being driven from below is that the tools entered the organisation through the consumer channel, not the procurement channel. They were free, they were accessible, and they produced results faster than any internal approval process could match. 70% of Australian leaders have already acknowledged this publicly, conceding their organisation lacks a plan and vision to keep up (Microsoft & LinkedIn, 2024). The acknowledgement is correct. The plan, where it exists, is mostly catching up to what has already happened.
Education, comms & training is the default response
What follows from the diagnosis is a strategic choice that most executives are making without knowing they are making it. Deloitte’s Australian survey of director-to-C-suite leaders finds that the most-cited response to the AI skills gap is educating the broader workforce. 53% of organisations are taking this path, while only thirty per cent are reimagining the organisation itself around the new patterns AI usage is producing (Deloitte Australia, 2026). Education fits inside the existing operating model. Reimagining changes it.
When the redesign is already underway from below, the choice to educate-only sits the executive in the wrong position. It treats AI as a capability gap inside the current shape of the work, when the shape of the work is itself the question. The cost compounds. Each cycle that passes with the structural decisions unmade accumulates more local adaptation: more workflows quietly reconfigured, more decision rights migrated, more handoffs absorbed. Each of those becomes harder to undo. The organisation does not arrive at a redesigned operating model; it arrives at an operating model that has been redesigned, by people without the line of sight to make the call. The cost is not in the productivity being captured. It is in the directions the organisation has now committed to without choosing them.
The decisions only you can make
The reorientation is to name the decisions that cannot be delegated to HR or IT, because they precede both. The strongest articulation of the leadership question right now belongs to McKinsey’s Kweilin Ellingrud, who frames it as whether organisations are prepared to redesign work itself around a new partnership between people and increasingly intelligent agents and robots, rather than automating tasks in isolation (Ellingrud, 2026). The redesign call belongs to the executive. Four decisions sit at that level and nowhere else.
Decision #1: What tools will you support
The boundary between AI tools the organisation will support and tools it will not is a question of risk appetite, sovereignty, and capital; it is not a procurement matter. Until that boundary is named, employees draw it themselves.
Decision #2: Governance and accountability
When AI is in the loop, accountability for the outputs shifts in ways the org chart was not designed to handle. The executive call is where accountability sits when a model makes a recommendation a human approves on the basis of a brief they did not write.
Decision #3: Funding training
The choice between funding training that fits people to current structures and funding the redesign of those structures is a choice about which version of the organisation the next investment is buying. Both can be funded. Neither will be funded enough if the choice is not made deliberately.
Decision #4: Decide what to 'stop doing'
The fourth is the call about what the organisation will stop doing. The AI work redesign is not additive. Some categories of activity become uneconomic. Some products become obsolete inside their own lifecycles. Some teams become structurally redundant before they realise it. Naming what stops is the most senior act of redesign, and it is the act the layers below cannot perform.
The executive who reads the system before the system finishes redesigning itself is operating eighteen months ahead of the one who reads it only after.
What this means for senior leaders
The AI work redesign in your organisation is not a question of whether; it is a question of who is making the call. If the executive has not named the four structural decisions — sanctioning, decision rights, capital allocation, and what to stop — they are being made by default at the layer closest to the keyboard.
Educating the workforce is a capital allocation choice, not a strategy. Funding training inside an unredesigned operating model commits the organisation to the shape it had before AI arrived. The reimagining of structures is the other half of the capital question, and it requires an explicit executive call.
The compounding cost of unmade decisions is greater than the productivity captured. Each cycle of unsanctioned adoption converts into structural commitment that becomes harder to unwind. The cost surfaces eighteen months later, in directions the organisation never deliberately chose.
The redesign call sits above HR and IT — they execute the brief; they do not write it. Treating AI as an HR upskilling problem or an IT procurement problem mistakes the level of the question. The brief is structural and belongs at the C-suite before it is delegated.
References
- Deloitte Australia. (2026). The state of AI in the enterprise — 2026 AI report. Deloitte. https://www.deloitte.com/au/en/issues/generative-ai/state-of-ai-in-enterprise.html
- Ellingrud, K. (2026, January). A new year’s resolution for leaders: Redesign work for people and AI. McKinsey Global Institute. https://www.mckinsey.com/mgi/media-center/a-new-years-resolution-for-leaders-redesign-work-for-people-and-ai
- Microsoft & LinkedIn. (2024, May 9). Work trend index 2024: AI at work is here. Now comes the hard part. Microsoft Australia News Centre. https://news.microsoft.com/en-au/features/ai-at-work-is-here-now-comes-the-hard-part/
