Beware offshoring dressed up as AI innovation and transformation

I have spent enough time around technology strategy to know that the language organisations use can tell you a lot about what is really going on.

When leaders start talking about “AI-enabled transformation”, “future-ready operating models”, “operational excellence”, and “global delivery”, my ears prick up. Sometimes those phrases describe genuine change. Sometimes they are a glossy wrapper around a much older story about cost cutting, outsourcing, and moving work to lower-cost jurisdictions.

That does not mean the technology is fake. It does not mean there is no modernisation happening. But it does mean we should be much more careful about accepting transformation rhetoric at face value.

I think this matters because words shape accountability. If a board, workforce, or customer is told that a major restructuring is about AI innovation, they are likely to assess it differently than if they were told plainly that it is also an offshoring or outsourcing program.

The Allianz example

The recent iTnews report on Allianz Technology is a good example of why I am wary. Allianz’s global Speed2Value initiative was presented in the language of innovation and operational excellence, but material examined in an Australian unfair-dismissal matter described it as a program to move work from high-cost jurisdictions into offshore or outsourced delivery models.

According to the report, Allianz Technology Australia’s direct employee headcount fell from 515 in March 2023 to 329 in March 2026. In the specific Fair Work Commission matter, work previously performed by an analyst programmer was transferred to India-based HCLTech, even though the redundancy was found to be genuine in the legal sense.

None of this means Allianz is unique. Many organisations have used offshore and outsourced technology models for years. The issue for me is not that offshoring exists. The issue is whether leaders are being honest about what kind of change is actually underway.

If the main business effect is lower labour cost through offshore delivery, then that should be said clearly. If AI is genuinely changing workflows, improving decision-making, or removing repetitive work, that should be evidenced separately. Rolling all of that together under the banner of “transformation” can make a very old management play look like a breakthrough.

Why this keeps happening

Generative AI has made this easier because it gives executives a very convenient storyline. AI is real. It is changing work. It is changing how organisations think about service delivery, content production, software development, and knowledge work.

But that broad truth can also be used to blur the specifics. A local team can be reduced, some tasks can be shifted to a vendor in a lower-cost country, a few AI tools can be added to the workflow, and suddenly the whole exercise is presented as an AI transformation. That may be partially true, but it is often not the full truth.

I think boards need to become much more disciplined here. They should be asking a few very plain questions:

• Which tasks have actually been automated, and what evidence shows that?
• Which tasks are still being done by people, but by a different workforce or vendor?
• How much of the projected saving comes from automation, and how much comes from labour arbitrage?
• What local capability is being lost in engineering, architecture, cyber security, product knowledge, and incident response?
• Who remains accountable when an AI system, an offshore provider, and a local business team each own part of the process?

Those are not abstract governance questions. They go directly to resilience, service quality, and organisational memory.

Other versions of the same story

This pattern is not limited to classic offshoring. Sometimes the language of AI efficiency is used to justify workforce changes, only for organisations to discover that the human work did not disappear at all.

Klarna is one of the clearest examples. The company became famous for claiming that its AI customer service assistant was doing the work of hundreds of agents. Later reporting showed that Klarna resumed hiring for customer support after its chief executive acknowledged that an excessive focus on cost had reduced quality, with the company looking at more flexible and freelance-style staffing arrangements. The work was still there. The labour model had changed.

Duolingo is another example worth noting. Reporting on its “AI-first” strategy said the company planned to phase out contractors where AI could do the work and expected teams to justify why a role could not be automated before asking for more headcount. At least that is direct. It tells staff and the market that workforce redesign is part of the strategy, rather than pretending the only story is product innovation.

I keep coming back to the same point: not every AI job story is really an AI story. Sometimes it is a contractor story. Sometimes it is a vendor story. Sometimes it is an offshoring story. Sometimes it is all three.

The capability question

This is where the Australian angle matters. When organisations reduce local technology capability, they are not just removing cost. They may also be removing context, institutional memory, and the practical ability to govern complex systems.

That matters even more with AI. Safe deployment is not just about buying a model or subscribing to a service. It requires people who understand the business process, the data, the regulatory environment, the security risks, and the real-world consequences when something goes wrong.

An offshore team can absolutely contribute to that work. I am not interested in simplistic arguments that treat local as automatically good and offshore as automatically bad. But I do think organisations need to be honest about whether they are retaining enough accountable in-house capability to challenge vendors, respond to incidents, and govern AI-enabled systems properly.

For Australia, this also connects to sovereignty. I have written before about sovereignty in the context of AI, data, and infrastructure. To me, sovereignty is not autarky. It is having meaningful choices, retaining critical capability, and avoiding a position where strategic dependence is mistaken for efficiency.

Call things what they are

I am not opposed to AI. I am not opposed to modernisation. I am not even opposed, in principle, to outsourcing or offshoring. There are times when each of those can make sense.

What I object to is managerial euphemism.

If work is being moved offshore, call it offshoring. If it is being handed to a vendor, call it outsourcing. If a task is genuinely being automated, show how and where that automation works. And if all three are happening at once, then boards, workers, and customers deserve a clear explanation of the mix.

Transformation should mean more than a new slide deck, a consulting narrative, or an AI label attached to an old cost-out playbook. It should mean the organisation is actually improving how it creates value, serves customers, manages risk, and retains the capability needed to govern its own future.

Anything less is not transformation. It is rebranding.