“Qantas Offshore 1,000 jobs!”.

AI was supposed to end offshoring. It seems to be doing the opposite.
Given yesterday’s announcement re. Qantas offshoring, it seems, most of our industry’s predictions about AI’s impact on offshoring has turned out wrong.
Qantas is reportedly in early discussions with Accenture about a program known internally as Project iQ, which could move up to 1,000 back office roles to India, including marketing, finance and HR. Qantas says no formal agreement is in place and no decisions have been made.

Clearly there’ll be an outcry from many circles, given the airline’s history with outsourcing, but this isn’t an isolated case. Woolworths began consulting staff in June about moving hundreds of finance, HR and IT roles to its hubs in India and the Philippines. NAB is hiring more than a thousand people across Vietnam and India. Officeworks is moving customer service roles offshore. Telstra signed a $700 million AI and data joint venture with Accenture.

Every one of these has been framed as an AI story. That’s what is surprising.

Two years ago the accepted wisdom was that AI would kill offshoring. Why send work to India when a model can do it for the price of an API call? We had that exact conversation about our own Nibble Edge studio. We sat around wondering whether we’d spent years building something with a short shelf life. Would all the love and energy put into the team be wasted?

Enjoying Mumbrella? Sign up for our free daily newsletter.

We were wrong.

AI runs on labour

AI doesn’t turn up as a finished product. It turns up as a capability that somebody has to build, feed and check.

Someone has to clean the data, connect the tools to the systems that already exist, write the prompts, test the outputs, spot where the model is confidently wrong, and build the checks that stop it happening twice. Then someone has to run the thing every day, at volume, to a standard.

That is skilled technical work and there is a lot of it. It’s why India, the Philippines and Vietnam are doing well out of this rather than being made redundant by it. Instead of replacing offshore teams, AI has given them more to do.
Agency land needs to calm down
Which brings me to the doom-saying, because our industry has been at it for two years now and it’s wrong. Don’t think, we haven’t fallen into the trap.

We thought AI would be the death of our DX division. Why would anyone pay an agency to build a website when a model can write the code? Instead we’re building sites in about a third of the time we used to. The work didn’t disappear. It got faster, so clients who could only afford one project a year can now afford three. Quicker prototyping, more tools, better experiences. Instead of worrying about the code or the tools behind all our projects, there’s more time for brand and strategy.

We thought AI would be the death of creative production. Instead we’re producing more. More assets, more variants, more personalisation, more of the tailoring that clients always wanted and could never justify paying for. Nobody has asked us for fewer assets since AI arrived. Not one client.

We assumed AI would remove the work. It removed some of the cost of the work. And demand did what demand always does when something gets cheaper.

So when I hear an agency describe AI as an existential threat, I’d gently suggest they’re not describing the technology. They’re describing a business model built on selling hours. If your value is the number of people on the account, then yes, this is frightening. If your value is the outcome, this is the best thing that has happened to us in a decade; more focus on making a business impact and proving our worth to the client.
We got lucky.
I want to be honest where we are now. We didn’t see this coming.

We’ve had a studio team in India for six years. We built it because we wanted to give clients quality creative production at a cost that made sense, not because we had any clever read on where AI was heading. Last year we announced a 24-hour studio and made the economic argument for what we called a “productivity partner”. At the time that was an argument about time zones, shifts and the cost of production.

AI has since made it a much better argument than the one we were making then.

This “big end of town” play isn’t available to most marketing teams.

A $700 million joint venture is not an option if you’re running a $1.5m marketing budget on $20m of revenue. Neither is a thousand role transition, a global capability hub, or an eighteen month transformation program with a consulting firm attached. But the underlying maths is the same, and arguably worse at that end of the market.

Look at where a mid-sized marketing budget actually goes. Not the strategy or the big idea. The rest of it. Versioning assets, adapting one campaign for six channels, building and trafficking media, updating the website, pulling reports together every month so somebody can ask why the numbers don’t match, and redoing jobs because the brief changed after production started.

That’s the work that quietly eats the budget. It’s also the work that responds best to good process, AI and people who do it every day at scale. And it’s almost always spread across three or four suppliers who don’t talk to each other. That is money wasted in the gaps between agencies, channels and teams.

Nibble Edge team in Bangalore
What we’re doing.
Rather than winding the studio back, we’re doubling down. We moved our team to Bangalore in April, and we’re now building an AI centre of excellence inside it.

Sitting over the top of it is the client facing team in Australia who own the outcome. As we all know, human direction always matters more than technology.

The point isn’t AI for its own sake. It’s putting AI into the parts of the process where it earns its keep: creative production and adaptation, media operations, ad ops, reporting and quality control. Same output, less time, less cost, fewer handovers.

So despite restless nights wondering whether Nibble Edge had made the right decisions as AI swept in, I’m starting to sleep easier.

The big companies have made their call on offshoring and AI. The more interesting question is what the rest of us do.