Mastercard & Rogers Bank test AI shopping checkout

Mastercard, Flybits and Rogers Bank have completed what they described as Canada's first agentic commerce transaction using an AI shopping assistant. The transaction involved a Rogers Red Mastercard cardholder.

A Flybits-powered assistant followed the customer's instructions to find an eligible product within a set budget and complete the purchase through Mastercard Agent Pay. The test was designed to show that an AI agent can complete a purchase within predefined spending limits, using checks on the consumer, credentials and merchant involved.

The initiative is built around a model in which the customer retains control of the transaction rather than handing decisions entirely to an autonomous system. The process was intended to give the cardholder visibility into preferences and guardrails while allowing the software to handle product selection and payment execution.

The transaction is one of the clearest examples yet in Canada of a payments network, card issuer and AI commerce platform working together on an end-to-end purchase. It comes as financial groups and technology companies try to define how AI tools can move beyond search and recommendation into actual payment activity.

Mastercard said the work builds on its Agent Pay framework, designed to support AI-driven transactions across existing payments infrastructure. The company has positioned the framework as a way to create rules and controls for software agents that may increasingly act on behalf of consumers.

Craig Reiff, senior vice president, core payments, Mastercard Canada, said trust will be central to broader adoption.

"Agentic commerce represents one of the most significant shifts in the future of commerce and payments," Reiff said. "As AI becomes increasingly capable of acting on behalf of consumers, trust will determine how quickly these experiences scale. Mastercard is helping establish the standards and infrastructure for secure, transparent and consumer-controlled commerce in Canada."

Consumer caution

The announcement was paired with survey data suggesting that consumer appetite for AI shopping remains measured. Mastercard said 41% of Canadians are open to using AI assistants to help them shop online, while 53% of those with positive views of AI shopping assistants are open to AI-assisted purchasing within preset preferences and budget limits.

Those figures point to a market that is curious about automated shopping but still sensitive to the conditions under which it would accept machine-led decisions. Here, the emphasis was on spending caps, validated instructions and a transaction flow that keeps the consumer involved in oversight.

Rogers Bank framed the exercise as a test of whether AI tools can meet the expectations customers already have for card payments. That includes maintaining familiar standards around security and clarity in how purchases are made.

"We are always looking for innovative ways to enhance the customer experience and deliver value to Canadians," said Nick Bednarz, CEO, Rogers Bank. "This collaboration shows how AI technology can simplify commerce while preserving the security, transparency and control consumers expect from their payment experiences."

Explainable layer

Flybits supplied the AI layer behind the transaction. Its platform was used not only to generate recommendations but also to show how and why those recommendations were made, reflecting a wider push in financial services for explainable AI rather than opaque decision-making.

That issue matters because agentic commerce raises questions beyond payment authorisation. If software agents are selecting products, comparing offers and finalising purchases, companies will need ways to show how those choices were made and whether they followed the user's stated interests and limits.

Dr. Hossein Rahnama, founder and CEO of Flybits and the Edward Rogers Sr. Chair in AI and Human Creativity at TMU, said transparency and relevance will be necessary if consumers are to accept such tools.

"For consumers to embrace AI-powered commerce, the experience must be transparent, fast and personalized," Rahnama said. "By combining contextual intelligence with auditable AI, we are building a new generation of digital commerce experiences that enable consumers to make more informed decisions with greater confidence, clarity and access to relevant choices."

The companies also pointed to permission-based data as an important part of the model. Mastercard said its Insight Tokens are intended to let AI agents use approved customer insights in a governed way, with consent controls built into how information is accessed and applied.

For banks, networks and merchants, the broader challenge will be turning one-off demonstrations into repeatable standards that work across sectors and platforms. The Canadian transaction offers a practical example of how that could look, with a consumer setting boundaries first and the AI agent acting only within those instructions.