Qantas will cut 850 Australian jobs and move some corporate functions to Accenture, including its India-based team, under a multi-year agreement to "modernise its back-end systems, processes and data." The first changes take effect by the middle of next year. Finance, procurement, human resources and marketing are the first targets.
That is the plain version. The version Qantas told is that this is an AI-driven overhaul. The gap between those two sentences is the story, and it has a direct consequence for anyone running a smaller company.
The news
The Guardian reported the cuts on 10 November. Qantas chief executive Vanessa Hudson said the "scale and speed of change in this space is unprecedented and unlike anything businesses have seen in generations," and that the decision was "not a reflection of their contribution or commitment."
Two days later, Reuters reported that Australian unions, a Greens senator and a University of Sydney professor are calling the announcement "AI washing": attributing layoffs to AI when the actual driver is cost-cutting and offshoring.
"If AI is making these jobs redundant, why does the work first need to be transferred to another group of employees?" asked Uri Gal of the University of Sydney Business School.
The Australian Services Union put it less politely, accusing Qantas of "using AI to launder their cuts to Australian jobs." The ACTU's Melissa Donnelly said unions were concerned about "companies using AI simply as a cover to cut Australian jobs with no accountability."
Qantas is the third major Australian corporate restructuring in three weeks, after Canva (1,200 roles) and National Australia Bank (2,500). About 4,550 corporate roles have been targeted, relocated or restructured in a month, with unemployment at 4.7%, a five-year high. The Reserve Bank of Australia said this week that AI has not yet caused widespread job losses, though it sees "tentative signs of adjustment."
Why it matters
The label is not cosmetic. It determines what you do next.
If AI genuinely did the work, your next move is to buy or build a tool. If the work moved to a cheaper team, your next move is to manage a vendor and a handover. If neither happened and the role was simply removed, you have a process problem that will resurface.
Most companies make the third mistake. They cut the person, keep the work, and discover in month two that the work was 30% task execution and 70% judgment, relationships and exception handling that was never written down.
You do not need 32,000 employees to repeat that mistake. A five-person firm can make it with one hire.
What to do about it
1. Classify the task, not the role. Before any headcount decision, list what the person actually does and sort it into three buckets.
| Bucket | Test | Typical examples |
|---|---|---|
| Automate | Same input, same output, no judgment | Data entry, invoice matching, scheduling confirmations |
| Delegate | Repeatable but needs a human on the other end | Inbox triage, client follow-up, vendor chasing |
| Keep | Judgment, negotiation, relationship, exceptions | Pricing calls, escalations, hiring, client strategy |
If most of the role lands in "keep," automation will not remove it. If most lands in "automate," you have a tooling decision, not a hiring decision.
2. Document before you cut. A handover file needs five things: inputs, tools and logins, decisions the person makes without asking, exceptions they escalate, and the weekly cadence of the work. Without that file, neither a tool nor an offshore team can absorb the role.
3. Ask three questions of any "AI did it" claim, whether it comes from a vendor, a consultant or your own team:
- Which specific tasks were automated, and what percentage of the role were they?
- What is the error rate, and who checks the output?
- Who handles the exceptions the tool cannot?
If those three cannot be answered with numbers, the AI claim is decoration.
4. Price the handover, not just the rate. A cheaper hourly rate with a three-month ramp and no documented process is usually more expensive than the role you replaced.
The Stealth take
We staff more than 1,200 assistants across 35+ industries, which gives us an unglamorous view of the same question. The roles that get easier to automate are the ones with a written process. The roles that survive are the ones where the value is in the exception handling.
The pattern we would flag for any owner watching this story is not "AI is coming for your assistant." It is that the companies most exposed are the ones that never wrote down what the role does. That is a documentation problem disguised as a technology problem, and it is cheaper to fix before the announcement than after.
Sources
- The Guardian, Qantas to cut 850 Australian jobs and outsource work to India under AI-driven overhaul, 10 November 2026.
- Reuters, Australian companies face 'AI washing' claims after wave of job cuts, offshoring, 12 November 2026.
Related: what a virtual assistant actually does · how to delegate before you hire