MCP, the new NDC? Why Africa's back-end systems will decide the agentic booking race
A significant shift is taking shape in travel distribution, and it carries a familiar set of risks for African businesses. Sabre has suggested that MCP could become the new NDC, pointing to nearly 80 customers already piloting or running the technology in production. Travelport and Amadeus are moving in the same direction, which means the industry is no longer debating whether agentic travel arrives, only how quickly.
For those unfamiliar with the terminology, MCP sits at the layer where artificial intelligence meets distribution. It allows AI agents to query systems, retrieve content and execute transactions directly, rather than a human consultant clicking through screens. The promise is speed, personalisation and vastly reduced friction. The assumption underneath it is that the systems being queried can actually respond.
That assumption deserves hard scrutiny in African and MENA markets. The uncomfortable question facing the region is what happens when the AI layer arrives before the infrastructure beneath it is ready.
Progress has certainly been made. A number of African carriers have invested seriously in NDC and can now distribute rich content, dynamic offers and ancillaries through modern channels. These are not token efforts, and the airlines concerned deserve credit for moving early. But they remain the exception rather than the standard. Much of the continent's airline distribution still runs on traditional GDS connections and fragmented connectivity, assembled over years through a patchwork of agreements, local aggregators and workarounds.
The deeper issue sits beyond the booking itself. Even where NDC capability exists, a transaction only becomes real revenue once it has been ticketed, serviced, refunded and settled. Those four steps are where African distribution most often strains. Servicing a changed itinerary, processing a refund across currencies, reconciling settlement between an airline, an agency and a payment provider in different regulatory environments: none of this is glamorous, and all of it determines whether a booking succeeds or collapses into a complaint.
This is why MCP may not solve Africa's distribution problem. It may expose it. An AI agent can only sell what the systems underneath it are able to fulfil. If an intelligent interface promises a seamless multi-airline itinerary with instant confirmation and automated refunds, but the underlying pipes cannot deliver, the technology simply makes the gap visible to customers who previously never saw it. Worse, it surfaces that failure at scale and at speed.
The practical implication for agencies, airlines and payment providers across the region is that the next distribution race will be decided by infrastructure, not interface. Investment in a polished AI front end delivers very little without parallel investment in reliable ticketing, robust servicing workflows, clean data and dependable settlement and payment rails. Payment providers in particular occupy a more strategic position than many realise, because agentic commerce assumes transactions clear instantly and predictably.
There is a genuine opportunity here alongside the risk. Markets that lack heavy legacy investment can sometimes move faster, skipping intermediate stages entirely. Mobile money across the continent demonstrated exactly that pattern. African carriers and agencies that treat this moment as a reason to modernise fulfilment rather than merely bolt on an AI layer could emerge with distribution capability that competes globally.
The businesses most at risk are those that adopt the interface and ignore the plumbing. The ones best positioned are those asking an unfashionable question right now: can our systems actually complete what an AI agent might promise on our behalf?
Agencies should be putting this directly to their airline partners and technology suppliers over the coming months, because the answers will shape competitive positioning for years rather than seasons.
