At Court 2 in the High Court Complex, Ring Road, Ibadan, Oyo State, a tired judge squints at a fraying docket while the courtroom groans under the weight of yet another overloaded case list. His fingers — tinged red from relentless notetaking — shuffle papers as he signals an abrupt end to the day’s proceedings. For many judges across Nigeria’s 36 states, this is a familiar scene: crammed calendars, dwindling patience, and the thankless labour of manually transcribing complex courtroom dialogue.
But in a country where the wheels of justice often grind slowly, a small but formidable startup is introducing a technological jolt to the system. Intron, a two-year-old Nigerian AI firm, is now actively powering speech-to-text automation in Nigerian courtrooms, starting with Ogun State. Its flagship voice AI suite, Sahara, built specifically for African accents, is being quietly adopted to replace weary scribes with real-time transcription and voice recognition tools designed for the linguistic complexity of the continent.
“We used to spend four hours on cases that now conclude in two or three,” reports the Office of the Chief Registrar at Ogun State High Court. “My Lord no longer writes during proceedings. He listens. He focuses. It’s been transformative.”
Founded in 2021 by Tobi Olatunji, a former physician, Intron has carved out a rare niche in Africa’s fragmented AI ecosystem: voice technology built by Africans, for Africans. With over 3.5 million audio clips collected from 18,000 speakers in more than 30 countries, Intron’s Sahara models recognize over 300 African accents and dialects, including Ghanaian English, Hausa, Zulu, Swahili, and even Arabic-English hybrids.
While global tech giants like Google and AWS continue to struggle with African names and pronunciations, Intron’s models are trained on precisely what their global counterparts often overlook — everyday African speech.
“We didn’t just tune a global model to local accents,” says Olatunji. “We started from scratch with real African data. Sahara is more than a technical win — it’s an ecosystem triumph.”
With this approach, the startup has achieved rare benchmark victories: outperforming OpenAI, Google, Microsoft Azure, and Amazon AWS on African speech recognition in recent public tests.
From Courtrooms to Clinics, Call Centers, and Chatbots
The courts are just one slice of Intron’s expanding footprint. In healthcare, Rwanda’s Ministry of Health is using Sahara to speed up documentation in its new electronic medical record system, while EHA Clinics in Abuja and Kano report a 60% reduction in time spent transcribing clinical notes.
In telecommunications and customer service, Branch International — one of Nigeria’s most prolific digital lenders — has embedded Intron’s conversational AI to deliver outbound engagement after hours. With Sahara CX Intelligence, users now get late-night customer support from voicebots indistinguishable from real agents.
In South Africa, nonprofit health platform Audere is experimenting with Sahara to power reproductive health chatbots for youth. In Uganda, the C-Care hospital network uses the tool to cut patient wait times and reduce errors across its 20 facilities.
Intron’s clients now include a growing list of government agencies, healthtech startups, and regional hospital associations, such as Helium Health (Nigeria), RUPHA (Kenya), Elephant Healthcare, and Rescue.co.
A New Class of African AI
At the heart of Intron’s rise is its Sahara product line — each model targeting a specific high-impact use case:
- Sahara-Optimus: Cross-domain speech recognition tuned to over 300 African accents.
- Sahara-TTS: The first continent-wide speech synthesis model, with 80+ lifelike voices in 40+ accents.
- Sahara-Voice-Lock: MFA and voice authentication built for African phonetics, with fraud prevention capabilities.
The company is now training the Sahara-Titan model — a next-generation speech-to-text engine capable of transcribing and translating in 20 of Africa’s most spoken languages. Meanwhile, Sahara-Primus is being designed to generate fluent, human-like speech in those same languages, unlocking better experiences for education, media, and content creators.
This advancement is fueled by a growing trove of local data: 30,000 hours of voice recordings across 64 African languages from over 32,000 speakers. As Olatunji puts it, “We built for the hardest environment first — crowded hospitals — and now our tech thrives everywhere.”
Despite the global AI boom, over 2 billion people in Africa remain underrepresented in the development of frontier technology. Poor recognition of African names, low data availability, and the continent’s linguistic complexity have made voice AI particularly challenging. Yet it is precisely in voice AI where Africa may leap ahead, with mobile-first populations, oral cultures, and soaring demand for AI-driven services in education, health, and governance.
Following a $1.6 million pre-seed round in 2024 led by Microtraction — with participation from Plug and Play, Octopus Ventures, Jaza Rift, and others — Intron has aggressively expanded its engineering and research teams, enabling both cloud-native and on-premise deployments across eight African countries.
The funding also drew interest from major global institutions like Google, NYU, Optum, and ClEAR, a testament to the technical seriousness of Intron’s vision.
“Intron represents a future where no community is left behind by technology,” says Olatunji. “Rather than complain about bias in global models, we built better ones.”
Intron’s rise is part of a broader wave of African AI innovation rooted in solving practical, high-impact challenges with local solutions. From Lagos to Nairobi, Accra to Kigali, a new generation of AI startups is emerging — not as copycats of Silicon Valley, but as pioneers of African-first design.
For Nigeria’s legal system, long trapped in the analog age, Intron’s courtroom debut may signal something rare: a path to justice that is faster, more efficient, and more inclusive. For African AI, it could be the moment the continent began building — and owning — its voice.