If there is a single thread running through South Africa’s start-up funding this year, it is not the sector the money has gone into. It is the customer the money is chasing. A pattern emerges from a run of recent rounds spanning workforce technology, payments, customer service, healthcare, logistics and fraud prevention: the businesses attracting capital are almost never selling to the person who ultimately benefits from the product. They are selling to the institution standing between that person and the market — the employer, the merchant, the bank, the clinic.
Jem, a Johannesburg-based earned-wage-access company, recently raised $8.4m in a Series A round led by Quona Capital in a deal that valued the business on the strength of its employer relationships rather than its worker base directly. The product lets employees draw down wages they have already earned, over WhatsApp, within 90 minutes. But the customer who signs the contract and pays the bill is the employer — more than 200 of them, including Servest, Edgars and KFC franchise operator Albimix. Jem reports net revenue retention of about 120 per cent and recurring revenue growth of more than 100 per cent year on year, figures that describe an enterprise software business, not a consumer lending one.
The same structure recurs at Happy Pay, a Cape Town buy-now-pay-later start-up that raised $5m earlier in the year in a seed round led by Partech. Rather than charging shoppers interest or fees, Happy Pay bills merchants for access to a network of high-intent buyers and improved checkout conversion. The company has signed up more than 600,000 registered users without becoming a lender to any of them in the conventional sense. Chief executive Wesley Billett has framed the model as making cash-flow flexibility free for consumers by having merchants fund it instead — a description that also applies, with the roles reassigned, to what Jem is doing with employers.
Cue, the Johannesburg and London-based customer service platform, follows the identical logic in a different market. Its AI agents resolve more than 60 per cent of customer conversations autonomously across more than 500 companies, but the entity paying for that capability is the business fielding the conversations, not the customer having them. In March this year, Orca, a Cape Town fraud-prevention start-up closed a $2.35m seed round led by Norrsken22, to sell to banks and telecommunications companies that need to detect fraud on mobile wallets and agent networks — the fraud victims themselves are never the customer.
Shiprazor and Breaze Delivery both sell logistics capability to merchants shipping goods, not to the people receiving them, and AI Diagnostics, the Cape Town healthtech company that raised roughly R85m ($5.2m) for its tuberculosis-screening stethoscope, sells to clinics and health systems rather than patients. Across most of these companies and several distinct industries, the pattern holds without exception: the paying customer is always an institution.
A coordination layer, not a new rail
A second, narrower pattern sits inside the first. Startups, such as Moment, NjiaPay and Shiprazor, are solving what is structurally the same problem in different markets. In South African payments, no single provider covers every card scheme, mobile money channel, bank transfer route and in-person payment method a merchant might need, so Moment and NjiaPay have each built a coordination layer that routes each transaction to whichever underlying provider performs best, rather than becoming a payment provider themselves. Moment, which raised $22m earlier this month in a Series A round backed by AlphaCode Venture Partners, General Catalyst and MultiChoice, targets large enterprise merchants. On the other hand, NjiaPay raised $2.1m in seed funding led by Newion two months after a $1.3m pre-seed round, and targets mid-market companies processing between R500,000 and R1m a month — a segment its founders argue is underserved by orchestration tools built for bigger clients.
Shiprazor applies the identical logic to logistics instead of payments. Rather than operating its own delivery fleet, the Cape Town company plugs into e-commerce platforms and routes each shipment across a network of more than 20 couriers based on cost, speed and service history. In each of the these cases, the start-up’s product is not the underlying service — payment processing, or a delivery van — but the software deciding which underlying service to use.
That makes the contrast with Breaze Delivery notable. In June, Breaze announced it raised $1.23m from the Vumela Enterprise Development Fund, a joint venture between FNB Business Banking and Edge Growth. The company addresses the same fragmented, unreliable SME delivery market that Shiprazor targets, but does so by operating its own driver network rather than aggregating other people’s. Two companies, one diagnosed market failure, two opposing structural answers — a sign that investors are backing the problem rather than any single theory of how to solve it.
Localisation as the underwriting argument
A third recurring feature is how these companies justify their existence to investors. Rather than positioning as a better version of an existing product, several argue that products built elsewhere structurally fail to work in South African conditions. Orca’s founders, both formerly of the payments company Stitch, built their fraud-detection tools after concluding that Western compliance software is not built to recognise the behaviour of mobile wallets, agent-led banking or the merchant networks common in South Africa — a case made more pointed by data showing a 600 per cent rise in fraud incidents in the country between 2018 and 2022, according to the Southern African Fraud Prevention Service.
AI Diagnostics makes a structurally similar argument about hardware rather than software. Its Ostium digital stethoscope, paired with acoustic AI software designed to flag tuberculosis risk from lung sounds, was engineered in South Africa specifically to withstand harsh clinic conditions — a distinction the company’s backers describe as a deliberate departure from medical devices designed, in their words, “from a distance.” The stakes are not abstract: the World Health Organization’s 2025 TB report recorded 249,000 new cases and an estimated 54,000 deaths in South Africa in 2024, with more than half of patients reporting no symptoms, undermining conventional screening.
Jem’s choice of WhatsApp as its primary interface follows a related but distinct logic — not a claim that Western tools misread local conditions, but a decision to meet frontline and deskless workers on infrastructure they already use rather than requiring a new app download.
Substituting for scarce expertise, not inventing new categories
Where artificial intelligence appears in this set of deals, it is deployed to substitute for a specific, named form of scarce expertise rather than to create a new product category. The same way Cue’s agents stand in for trained customer service staff is almost the same way AI Diagnostics’ stethoscope stands in for a specialist clinician or an X-ray machine that many South African clinics do not have on site. In both cases, the pitch to investors rests on a shortage that already exists, not on a capability that did not exist before.
What the pattern suggests
When these are all factored in, the deals describe a narrower investment thesis than the “fintech, healthtech, logistics” sector labels suggest on their own. South African investors backing early and growth-stage start-ups this year have concentrated on companies that sell coordination, compliance or automation to an institution, using that institution as the route to a population — workers, patients, shoppers, the underbanked — that would be expensive or difficult to reach directly. The credit-avoidance structures at Jem and Happy Pay, the routing logic at Moment, NjiaPay and Shiprazor, and the localisation argument at Orca and AI Diagnostics are different tactics aimed at the same target: businesses that already have a relationship with the end population in question.
The overlap among backers also suggests some funds are placing multiple bets on the same structural thesis. Norrsken22 invested in both Shiprazor and Orca. E4E participated in Jem and Shiprazor. University Technology Fund appears in Jem and Happy Pay. Futuregrowth Asset Management is present in Jem through the Next176/FutureGrowth vehicle and in Happy Pay.
Nivesh Pather, investment principal at Norrsken22, said Africa’s e-commerce market “remains fragmented and unoptimised,” creating a need for infrastructure that helps merchants meet rising consumer expectations. That suggests some investors are deliberately backing companies that coordinate existing providers rather than replace them.
It is not a universal description of South African venture activity — larger, capital-intensive infrastructure plays elsewhere on the continent, in electric mobility and autonomous vehicle logistics, follow a different model built on owning physical assets rather than coordinating access to them. But within South Africa specifically, on the evidence of this year’s rounds, the businesses drawing investor interest are consistently the ones standing between an institution and the population it already serves.