0%
Loading ...

Meet the 17 Winners of the Sahara CodeSwitch Africa Challenge

At the Sahara CodeSwitch Africa Challenge, teams from 17 countries submitted over 100 voice AI solutions and benchmark reports, with winners across health, fintech, agriculture, education and public services.

Most Africans do not speak one language at a time. A trader in Lagos moves between Yoruba, Pidgin and English in a single sentence. A nurse in Nairobi switches between Swahili and English mid-consultation. Most global voice AI systems still break when that happens.

The Sahara CodeSwitch Africa Challenge, run by African voice AI company Intron, asked African builders to fix that. Starting with a 3-day hackathon at the 2026 Deep Learning Indaba in Lagos and followed by the main challenge in September, the task was to build a voice agent that completes a real task based on natural African mixed-language speech input, comparing Intron’s Sahara models against at least two other speech models. This comparison produced a benchmark report showing which models work best. The competition has now closed, with 17 winners named across five categories and three mission prizes.

Who entered the Sahara CodeSwitch Africa Challenge

The challenge drew 881 registrations, made up of 231 teams and 650 individual builders, from 45 countries. Forty of those countries are in Africa. Nigeria accounted for 644 registrations, followed by Kenya (46), Uganda (21), South Africa (19) and Ethiopia (16).

Interest was spread across sectors. Agriculture and education drew 26% of registrations, fintech and telco 21%, health 21%, and legal and public services 7%. The remaining 25% pitched other high-impact use cases.

126 submissions and 31 new test sets

After a month-long build period that ended on September 15, 126 teams from 17 countries submitted working solutions along with benchmark reports analyzing model performance in each sector. Thirty-one teams also contributed their own code-switched audio test sets, adding new evaluation data for African languages that are poorly represented in global AI.

“Code-switching is how Africans actually speak, and these teams showed what is possible when voice AI is built for that reality,” said Tobi Olatunji, CEO of Intron. “The insights from the benchmark reports and datasets they produced are as valuable as the solutions submitted. We’re truly impressed by Africa’s breadth of technical talent.”

Meet the 17 Sahara CodeSwitch Africa Challenge winners

Fintech, Telco and Call Centre

In Fintech, Telco and Call Centre, David Agbolade (United Kingdom) took first place with SautiLedger, a voice ledger for market traders who mix languages. Yusuf Anigilaje’s team ACYM (Nigeria) came second with Noba, a voice agent for bank fraud prevention and telecom customer care. Aderemi Ademola’s team Rave (Nigeria) placed third with Anuva.

 

Health

The Health Category had five winners across Africa, all focused on maternal health, voice triage, or clinical decision support:

  • Dawa Health (Zambia), led by Tafadzwa K Munzwa, with DawaMom, an agentic maternal and women’s health platform
  • Sauti Yetu (Kenya), led by Kelvin Ndambuki, with a code-switching voice triage tool for health facilities
  • Faulu (Rwanda), led by Bellah Ellam, with MAMA-SAUTI, built for Kenya’s community health promoters enabling voice-based screening on postnatal visits
  • Symbiosis (Ethiopia), led by Gebrehiwot Teklehaymanot Abraha, with Voice Triage Ethiopia, offline-first Amharic-English clinical decision support for Community Health Workers
  • Olumodeji Bolu (Nigeria), with Careline, a voice-first care navigator for expectant mothers

Agriculture, Education and Government Services

In Agriculture, Education and Government Services, Spyder Engineering Team (Nigeria), led by Agoro Oluwatimilehin Teslim, won with SautiCivic Bridge, a voice-first platform for reporting infrastructure failures or seeking legal assistance. Deborah Falope’s AquaTrack (Nigeria) came second with a voice tool for fish farm operations. John Fiewor (Nigeria) placed third with GradrAI Viva, an edutech solution that verifies students’ understanding through code-switched oral assessments.

Mission prizes

Three mission prizes rewarded specific goals:

  • Agentic Prize: Chukwuebuka Chukwudi (Nigeria) for iShop, a voice-powered Shopify shopping assistant; David Bakare’s Aalto (Nigeria) for voice access to civic forms; and Olivier Kamali (DRC) for Djema Voice, built for informal vendors and job seekers in Bunia
  • Language and Culture Prize: Henry Ezeanyim (Nigeria) for Nketa, a voice-powered Igbo cultural learning app
  • Masakhane Benchmark Dataset Prize: Team Yeha (Ethiopia), led by Welela Bekele Regassa, for Mama Health, and David Uje Nelson’s Weha Health (Nigeria), both voice-first maternal health tools

Sahara CodeSwitch Africa Challenge judges and partners

Submissions to the Sahara CodeSwitch Africa Challenge were reviewed by judges from organisations including Microsoft, CcHUB, Branch, ARM Investments, Gooey.AI and the Liverpool School of Tropical Medicine. The challenge was supported by partners including NVIDIA, CLEAR Global, Masakhane African Language Hub, Wema Bank, Branch, ARM, Gooey.AI, and CcHUB.

The winning teams now join a growing group of African builders proving that voice AI for the continent does not have to wait for global labs. With over 100 benchmark reports and 30+ new test sets on record, the challenge has also left behind something the field has long lacked: evidence of how speech models actually perform on the way Africans talk.

Facebook
Twitter

Other articles

The Future of African Voice AI

Gain deep insights into the evolving technological landscape with our upcoming 2026 Africa Voice AI Report. The report explores the trends, challenges, and opportunities shaping the future of speech interfaces across the continent.