Kenyan Startup Fastagger Builds Offline AI for Small Businesses on Low-Cost Phones

The assistant, called Auni, launched in 2024 and grew quickly once it was integrated into the business app of M-Pesa, Kenya's largest mobile money service. About 6,000 businesses have signed up since then, according to the report. Users include barbershops, restaurants and small manufacturers. Auni reads their transaction statements, charts sales patterns, identifies regular customers and suggests marketing steps.

Co-founder Mutembei Kariuki said he started thinking about AI for Kenya in 2019, at a University of Tokyo conference where he heard a lot about the technology's global impact. "But nobody mentioned Africa at all," he told IEEE Spectrum. He started Fastagger with Jude Mwenda and Stephanie Njerenga.

The founders went after the data problem first. Because more than 80 percent of Kenyan adults use mobile money apps, the team adapted an open-source optical character recognition model to read the apps' statements. Existing models struggled with these low-quality PDFs, which had unfamiliar layouts and sometimes stamps or errors. Reading statements, organising the data into tables and drawing charts can all happen on the phone without a network connection. Deeper analysis and answering questions about the data still need the internet.

Peter Chege, who runs a salon and barbershop in Nairobi, said he used to record sales on paper. Auni showed him that mornings were his slowest hours, so he now runs errands then and keeps fewer barbers in the shop until business picks up around 1 p.m. A Nairobi fish restaurant uses the tool to find its top customers automatically and offer them discounts. Before, that meant going through SMS receipts one by one.

Daniel Björkegren, a Columbia University economist who studies AI in developing economies, said the widespread use of basic phones in sub-Saharan Africa points to different ways of using AI. "The history of technology should suggest that this is not going to be people running Claude Code or Codex on their laptop," he said.

Fastagger has had support from accelerator programmes run by Google, Microsoft and NVIDIA. Its only employees are still the three founders, the report says.

The company's approach shows a different route to AI adoption in emerging markets. Instead of relying on large cloud models, it builds small models around mobile money data that already exists, which could make it a useful example for payment providers and investors in markets where most people do not have smartphones.