The payments industry is obsessed with speed, and it should be, but the real disruption arriving in 2026 is not a faster payment, it is a different payer. AI agents are moving from recommending what to buy to taking action on your behalf, searching, comparing, selecting, and paying. Commentators are already naming agentic commerce as one of the defining payments trends for 2026, alongside the rails and standards needed to make it work safely.
Most people still think of payments as a moment at the end of a journey, tap, click, confirm. That mental model made sense when commerce was human paced and decisions were made one by one. Agentic commerce breaks it. When an AI agent can execute a transaction, the payment is no longer just a transfer of value, it becomes a test of delegated authority.
A useful way to see it is this, the heart of payments shifts from moving money to proving permission. The question changes from, “Did the credentials work?” to, “Was the agent allowed to do this, for this merchant, for this category, within this price, under this context, and can we prove it?” That is why major networks are talking less about shiny features and more about infrastructure, standards, and interoperability as the defining work of the next era.
Agentic commerce forces four awkward conversations that the industry has postponed for years.
This is why the hottest topic is not “AI shopping” as a novelty, it is trust architecture for delegated transactions. Mastercard’s own forward looking framing of 2026 payments trends, including agentic commerce and AI driven security, is effectively an admission that trust and automation are converging into the same product challenge.
Here is the narrative that needs to die, “We will innovate first, then add guardrails later.” That approach is how fraud scales faster than adoption, and it is how regulators lose patience.
If AI agents become meaningful actors in commerce, the first major scandal will not look like a traditional breach. It will look like delegated harm, a customer saying, “My agent was authorised, but it was tricked”, and the ecosystem arguing publicly about whether the customer is to blame. In a trust industry, that debate is poison. The industry should treat agentic commerce the way aviation treats autopilot, autonomy is only acceptable when safety systems are at least as mature as the automation.
The policy implication is clear. Standards for consent, proof of payee, and dispute pathways should be built early, not after the fraud wave arrives. Europe’s instant payments regulation is already embedding ideas like Verification of Payee into mainstream payment requirements, a recognition that speed must be paired with preventative checks. Agentic commerce will demand the same mindset, but even more so, because the “actor” is software.
Agentic commerce is arriving at the same time as another structural shift, tokenised settlement moving from the edge into the centre. Stablecoins in particular are being pulled into mainstream payment infrastructure, not as a cultural statement, but as a settlement tool for faster, cheaper movement of value in specific contexts, especially cross border. The signal is not social media hype, it is institutional capital. A recent example, Mastercard’s move to acquire a stablecoin infrastructure firm, framed explicitly as a strategic bet on integrating stablecoin rails into payment networks.
Why does this matter for agentic commerce. Because autonomous systems thrive on rails that are programmable, instant, and interoperable. Stablecoin infrastructure, where appropriately regulated and controlled, starts to look like a compatibility layer for certain flows, particularly B2B and cross border, while instant account to account rails modernise domestic commerce. In other words, the rails are evolving to match the tempo of machine mediated transactions.
Africa’s payments story has always been less about perfect infrastructure and more about adoption at the edge. Mobile money grew because it solved distribution and daily usefulness, and it did so at scale. GSMA notes that mobile money reached major milestones in 2024, including more than 2 billion registered accounts globally and over half a billion monthly active users, with continued growth in transactions per active user.
The relevance to agentic commerce is not superficial. Africa has trained the world on a core lesson, people adopt systems that are simple, ubiquitous, and trusted, even if the underlying rails are complex. As instant payment systems proliferate across the continent, and as mobile wallets evolve into multi service ecosystems, the foundation for machine mediated commerce strengthens, not because everyone will suddenly deploy shopping agents, but because the rails become capable of supporting low friction automation.
There is also a warning. Faster rails compress the reaction window for scams. If the next era is about automated commerce, then automated fraud will follow. Africa’s next leap will not be won by the fastest payment, it will be won by the safest payment at speed.
South Africa is a particularly important test market because it combines sophisticated financial institutions with a strong push to modernise payment infrastructure. PayShap launched in March 2023 as part of the move towards instant, real time payments, explicitly framed as supporting modernisation, interoperability, and innovation. Independent ecosystem analysis also notes that PayShap supports P2P and some early P2M use cases, with broader merchant payment capabilities planned.
At the same time, South Africa is inching towards the other ingredient agentic commerce requires, a clearer framework for permissioned data sharing and payment initiation. The FSCA’s open finance programme, including its published materials and consultation documentation, signals that the regulator sees open finance as consent based data sharing and payment initiation under supervision.
Put those together and you get a simple strategic truth. If the rails are becoming instant, and the system is moving towards permissioned initiation, then South Africa is building the conditions in which agentic commerce could become real, but only if it is paired with a strong trust layer, identity signals, beneficiary verification norms, and clear liability pathways.
The hottest topic in payments is not AI, not stablecoins, not instant transfers. It is the emergence of payments as a trust and permissioning utility for automated actors. Agentic commerce is simply the most visible expression of that shift.
If you are a bank or PSP, your product roadmap should include permission controls that customers can understand, and that machines can enforce. If you are a retailer, assume that “discovery” will shift from humans browsing to agents selecting, and invest in machine readable offers, identity confidence, and frictionless dispute resolution. If you are a regulator, treat delegated payments as a consumer protection frontier, define accountability early, and make safety standards part of the rails, not an optional overlay.
The future direction is not a world where humans disappear from commerce. It is a world where humans delegate more, and expect more, faster, safer, and more reversible when things go wrong. Payments will be the industry that either enables that trust, or becomes the bottleneck that slows it.