Mastercard has completed a series of agent-led business transactions across Australia and New Zealand, testing how artificial intelligence (AI) could move beyond recommending actions to carrying them out.
While earlier experiments this year focused on consumer purchases, this latest round focuses on commercial payments and the day-to-day decisions small businesses deal with. The transactions were completed with a mix of banks (CBA, Westpac, Bendigo, HSBC and AMP), as well as merchants and platforms. These included the likes of Hnry, MYOB, and Pay.com.au.
Related Article Block Placeholder
Article ID: 334821
According to Mastercard, the aim is to link steps that are currently separate, where business data feeds into an AI system, generates a recommendation and then completes the payment. The distinction is that these transactions combine insight and execution, rather than leaving business owners to act on recommendations themselves.
“There’s still that disconnect… you have to come out of the AI platform to then go to that merchant to pay,” said Anouska Ladds, executive vice president of commercial and new payment flows for Asia Pacific at Mastercard, to SmartCompany. “That last mile is where agentic will play a role.”
What this means for small businesses
For small businesses, the relevance comes down to time and how work gets done.
Smarter business news. Straight to your inbox.
For startup founders, small businesses and leaders. Build sharper instincts and better strategy by learning from Australia’s smartest business minds. Sign up for free.
By continuing, you agree to our Terms & Conditions and Privacy Policy.
Mastercard says SMEs are spending around 60 hours a week running their business, with another 10 to 20 hours on administrative tasks such as payments and financial management.
“SMEs… don’t have a passion to be a CFO, right? So actually, where you can take that away and help them enable that… you’re giving the SME time back,” Ladds said to SmartCompany.
The demonstrations from Mastercard highlighted this focus on day-to-day financial admin. In one example, a business owner reviewing upcoming invoices was prompted to consider how and when to pay, based on cash flow and potential rewards. In another, an agent identified low stock levels and triggered a purchase from an existing supplier.
Related Article Block Placeholder
Article ID: 330566
In a separate demonstration, an agent surfaced upcoming bills and calculated how to maximise points by shifting payments onto a card, before completing the transaction.
According to Surin Fernando, senior vice president at Mastercard Australia, the aim is to reduce the number of steps involved between identifying a task and completing it.
“The piece that we’re most excited about is when you can bring the blend of that data with AI alongside facilitating an action that may potentially be even a payment,” Fernando said to SmartCompany.
The constraints behind the Mastercard pilot
That framing makes sense, particularly for businesses already juggling multiple systems. But the transactions shown here were completed in controlled conditions, with banks, merchants and software platforms working together.
“This is an ecosystem play… no one player is going to be able to do this in isolation,” Ladds said.
That reliance on multiple parties shapes how quickly this could move beyond testing. Banks need to support authentication and payment flows, software platforms need to integrate AI into existing products, and merchants need to make inventory and pricing data accessible in a structured way so that agents can act on it.
Related Article Block Placeholder
Article ID: 314553
The system also depends on underlying AI models and platforms that Mastercard does not control, adding another layer of dependency for businesses.
Fernando said there are still steps required across the value chain, including upgrades on the acquiring side and greater readiness from merchants.
It also raises the question of how these systems are surfaced to businesses. Rather than sitting in a single product, the agent experience could appear inside accounting software, banking apps or other tools SMEs already use, depending on how partners choose to implement it.
It is also unclear what the cost of these capabilities will look like for smaller businesses as they move beyond pilot stages.
Even so, Mastercard is positioning this technology as relatively close.
“It’s definitely a near-term [opportunity]… we’re not talking two to three years,” Ladds said.
But for now, the technology remains in testing, with wider rollout dependent on how quickly those pieces come together. It also raises practical questions around liability and errors, particularly if a system makes a decision that is technically correct but commercially wrong.