The global travel payments industry is having an urgent conversation about agentic commerce. At the Airline and Travel Payments Mega Event in Dublin this month, the central question was how payment systems authenticate, authorize, and assign liability when an AI agent books travel autonomously on behalf of a traveler. Visa has launched its Agentic Ready programme, now expanding into Asia Pacific and Latin America. Sabre, PayPal, and MindTrip announced what they describe as the travel industry’s first end-to-end agentic booking system, connecting GDS inventory, conversational AI, and payment into a single pipeline.
The industry is taking this seriously because the commercial logic is compelling. Boston Consulting Group estimated in late 2025 that AI agent-led shopping could represent more than a quarter of global ecommerce spending within the next several years. Major travel technology providers are no longer debating whether AI agents will become a booking interface. They are debating how to authenticate them, how to assign liability when they make errors, and how to build payment authorization frameworks that work at machine speed. The conversation has moved from experimentation to infrastructure.
These are consequential developments. But the payment infrastructure being designed for this agentic future is being built around card networks, digital wallets, and regulatory frameworks that reflect the markets the industry has always prioritized. Africa’s mobile money rails, which collectively moved $1.4 trillion across sub-Saharan Africa in 2025 alone, are not part of that conversation.
The default assumption in most coverage of this gap is that Africa’s infrastructure is not ready. That assumption deserves scrutiny. In several important respects, it is wrong. And the consequences of leaving it unchallenged will fall on the agentic commerce platforms, not on Africa.
The Infrastructure Is More Capable Than the Narrative Suggests
Africa’s mobile money ecosystem is routinely described in travel and payments media as nascent, fragmented, or developing. The numbers tell a different story.
MTN MoMo operates across approximately 16 markets with a unified API surface. That level of multi-market consistency is not something card networks achieved easily or quickly in their own expansion phases. M-Pesa supports both merchant-initiated and customer-initiated payment flows, with a documented API that handles pull payments, B2C disbursements, and reconciliation at scale. Interoperability agreements across East Africa now allow seamless transfers between M-Pesa, MTN MoMo, Airtel Money, and Tigo Pesa users in Kenya, Uganda, Tanzania, and Rwanda. The East African Payment System is extending this further.
The question is not whether mobile money can process transactions at scale. It already does. The question is whether its authorization layer evolves quickly enough to support machine-initiated commerce. That is a precise and solvable problem. It is not evidence of inadequate infrastructure.
Safaricom’s fraud detection on M-Pesa, including biometric authentication and real-time SIM-swap monitoring, is sophisticated by any standard. Card network gaps in agentic readiness are treated as engineering problems to solve. Mobile money gaps are treated as market limitations to work around later. That asymmetry is a choice, not a technical verdict.
Africa Has an Advantage Nobody Is Naming
There is a structural argument that Africa’s mobile money ecosystem is actually better positioned for agentic commerce than card-based markets, and it is almost entirely absent from the industry conversation.
Card-based payment systems carry decades of legacy. Magnetic stripes, static card numbers, and human-facing authentication flows were designed for a world of physical transactions. The move toward tokenization, 3DS, and now agent-compatible payment standards is a process of retrofitting infrastructure that was never designed for machine-initiated commerce. Every step forward requires negotiating with existing schemes, issuers, acquirers, and processors who have commercial interests in the current architecture.
Mobile money was built API-first, mobile-native, and without the physical card layer. Its authentication flows are already abstracted from hardware. Its transaction initiation is already software-driven. The leap from human-initiated to agent-initiated payment is architecturally shorter on mobile money rails than it is on card rails, if anyone bothers to build the bridge.
The constraint is not technical readiness. It is commercial prioritization. The companies building agentic commerce infrastructure are building for the users they know, in the markets they have always served. Africa is not in the room when those decisions are made.
The Counterargument, and Why It Does Not Hold
Critics would argue that mobile money remains nationally fragmented and that agentic commerce requires globally accepted payment credentials. That concern is legitimate. An AI agent booking a multi-leg itinerary across three continents needs payment infrastructure that works across jurisdictions, and no single mobile money operator offers that today.
But interoperability challenges are commercial and regulatory problems, not evidence of inadequate infrastructure. Card networks took decades and significant regulatory intervention to achieve cross-border interoperability. Mobile money interoperability across East Africa, which seemed equally intractable five years ago, is now functional across four markets and expanding. The GSMA Open Gateway initiative is creating programmable interfaces over mobile network assets that point toward regional and eventually global connectivity. The trajectory is clear. The argument that mobile money cannot support agentic commerce because it is fragmented today would have applied equally to card infrastructure before the global interoperability frameworks that now seem inevitable.
What This Means for Travel Distribution
For airlines, OTAs, and distribution platforms with serious African market ambitions, the stakes are direct. The NDC transition has already shifted the commercial logic of airline retailing toward direct payment relationships between carriers and travelers. In that world, the payment method a traveler uses is no longer abstracted by a GDS settlement layer. It is a live commercial question.
If agentic booking becomes the dominant interface for travel purchase over the next three to five years, and if the payment rails supporting it are built exclusively for card-holding consumers, the African traveler is not excluded from agentic commerce because her infrastructure is inadequate. She is excluded because the companies building the standards did not include her infrastructure in scope.
That distinction matters commercially. Africa’s middle-class traveler base is growing. Intra-African aviation is expanding. The Nigerian, Kenyan, Ethiopian, and Ghanaian travel markets represent real and increasing revenue. A distribution platform that cannot serve those travelers through an agentic interface will lose ground to one that can, once the bridge is built. And the bridge will be built, because the commercial incentive is large enough.
The question is whether the platforms shaping agentic commerce standards today build that compatibility in from the start, or retrofit it later at higher cost and lower quality.
The Consent Layer Is the Real Work
To be precise about what does need to be built: the gap is not in payment rails. It is in the consent and authorization architecture that sits above them.
Agentic commerce requires a user to delegate payment authority to an AI agent within defined parameters: a spending limit, a time window, a category of permitted purchases. On card networks, mechanisms along these lines, including stored credentials, network tokenization, and 3DS delegated authentication, are being extended toward agent use cases, however imperfectly. On mobile money rails, equivalent frameworks do not yet exist.
Building them requires coordination between mobile money operators, regulators, and the travel technology companies that want to access these markets. African central banks have not yet issued guidance on AI-initiated payments. Liability frameworks for agent-authorized transactions are unresolved. These are real gaps, but they are regulatory and commercial gaps, not infrastructure gaps. They are solvable with the same kind of industry coordination that produced mobile money interoperability across East Africa, which itself was a harder problem five years ago than agent authorization is today.
Who Needs to Act
The mobile money operators with the most to gain are the ones with the clearest mandate to move. MTN MoMo’s pan-African API surface and market reach make it the most obvious candidate to develop an agent-compatible authorization layer. M-Pesa’s technical maturity and merchant payment infrastructure give Safaricom and Vodacom a foundation to build on. An operator that develops a reliable, consent-compliant, agent-compatible interface over its rails will own a critical piece of African travel distribution infrastructure for the next decade.
The travel technology companies building agentic systems need to treat African mobile money integration as a launch requirement, not a roadmap item. The Sabre-PayPal-MindTrip partnership, Visa’s Agentic Ready expansion, and the payment standards emerging from events like ATPS are all being designed without African rails in scope. That is a choice, and it is one that will cost those platforms market share in the fastest-growing travel regions in the world.
The clearest opportunity sits with the fintech infrastructure players already operating across African mobile money rails. Cellulant, which connects merchants to mobile money networks across 18 African markets, Flutterwave, which supports payments across 34 African countries, and Onafriq (formerly MFS Africa), whose network connects over 500 million mobile wallets across 40 African countries, are each positioned to build the abstraction layer that connects agentic commerce stacks to M-Pesa, MTN MoMo, and Airtel Money. That is not charity work. It is a defensible and commercially significant infrastructure position, and the window to claim it is open now.
The Framing Needs to Change
The travel industry has a habit of describing African markets as catching up to a standard set elsewhere. In payments, that framing has never been entirely accurate, and in the agentic commerce era it is actively misleading. Africa did not need to replicate card infrastructure because it built something better suited to its own context. The question now is whether the global travel payments industry will treat that infrastructure as a serious input into the agentic future it is building, or whether it will repeat the same mistake it made with NDC: designing for the markets it knows, and leaving the rest to figure it out later.
The infrastructure exists. The commercial case is clear. What is needed is the decision to include it.
Travel Distribution News covers airline distribution, NDC, GDS dynamics, travel payments, and travel technology with a focus on Africa and emerging markets.
