Share this
Written By
Folake Ajao
Ehinome Ogbeide had spent years in the hit-record game before deciding to build something for four-year-olds. He started in events, moved into digital music distribution, and later led the data side at Mavin, where he worked on Rema's "Calm Down" (the first solo Afrobeats artist song to reach one billion streams on Spotify). He also played a major role in the global streaming success of Ayra Starr's "Rush," which earned a Grammy nomination. He is now co-founder of Bina AI, a co-learning platform that uses AI and African music to create songs and stories to teach children aged 4 to 12.
He was very practical about how he planned to build this. The events required him to be there constantly, to be talking to people, but he wanted to create something that could survive without that, and distribution was the way to do it. At Jungle's Distribution circa 2016/2017, he got a real grounding in catalog structure and DDEX and what it actually means to be pushing 7 or 8 releases a day. The job also taught him to plan for failure. "You have to be thinking about what can go wrong," he says. An agency called Fishbyte showed him how to get clients and price his work. Mavin then employed him as a Distribution Specialist. He's now the head of the team at Rahdah that builds products for clients and Rahdah's own initiatives.
Bina basically started at home. Ogbeide is the youngest of three, so small children were new to him until his niece moved in with him, and then other kids started coming over almost every day. He created a children's coloring book, started studying AI, and combined the two.
As he sees it, the problem has two major parts. There are no datasets for African children, so they grow up without any models built for them) and a large number of the most successful YouTube channels on YouTube are built on Western sounds. "Music is very native," Ogbeide says. "It's conceptual." He cites a viral video of a little kid who would dance when a Davido album played and completely ignore all the other music. He argues that the child is responding from the gut. He's currently researching this with a South African university and students at the University of Benin. The other element is personalization. YouTube isn't a curated experience for the child in front of you, and AI enables personalization and speed.
A parent opens Bina and types or speaks what the child is having trouble with (e.g., the alphabet). It then asks for the child's age, which determines the pace (a 4-year-old needs something much slower). It asks for the country, which determines the sound (Nigeria = Afrobeat, South Africa = Amapiano, Congo = Rumba). It asks for the name, so a song for Tolu might begin "Tolu, this is your teacher, repeat after me." The result is a 2- or 3-minute song that parents can download and play over and over. "You get a song, but it's a learning tool," Ogbeide says. It has a storytelling feature that writes and reads a bedtime story with the child's name in it, plus a community tab where parents can post what they've taught their kids. Ogbeide grew up in a world where everyone looked after everyone else, and he wanted to digitize that.
Ogbeide recorded the voices in a studio with real kids, and the AI learned what a child's voice sounds like in Afrobeat, Amapiano, and Rumba (the datasets Bina started with). Kenya and the other regions came next, since there was no dataset to begin with. He created one. A contact from his film work knew four or five people, one of whom trains kids at his church, and an existing children's choir also contributed. They sorted the publishing rights, and everyone was paid. "You can't do it all on AI because these are real kids," he says. The kids are very much aware of the tension and don't care about deadlines, so no one is shouting in that studio. There's a chat, and they say, "Right, take it again."
Testing took place in Lagos classrooms with 10 to 15 pupils at a time, from the very bright to the very slow. The most striking thing was punctuation. The storytelling text had absolutely no punctuation at all, and the team found it perfectly readable. The kids, however, couldn't read it at all as their brains hadn't been trained to insert the pauses. The moment we added punctuation, it was a completely different story. Focus was the real challenge. One of the kids would run past and end the session, so Ogbeide brought in his sister-in-law (a trained psychologist and educator the kids really liked) to help. The school moved visits from the end of the day to midday, when the kids were less tired and more animated, to keep them engaged.
Safety was at the heart of the design, so a parent or carer can sit with the child and manage risk that way. "We train our own models so what goes in is under our control," says Ogbeide. Children are represented as symbols (a lion's face, for instance) rather than photographs. We encode names, codes, and community on the backend, and no one can ever open another child's profile. The 4-12 age group follows the same principle: old enough to use a computer, but young enough that they can't use it without a parent/carer present. Parents have been far more receptive than he'd anticipated. Some still worry AI will corrupt their children; others are asking how their kids can start coding.
The product emerged from the AI & African Music Project at the University of the Witwatersrand in Johannesburg (run by the Wits Innovation Center and the Machine Intelligence and Neural Discovery (MIND) Institute). The project matched Ogbeide with an engineer in the DRC. They built Bina in roughly five months without ever meeting in person, with Ogbeide testing in Lagos and his partner testing in the DRC. They demonstrated it in Johannesburg in April 2026 to researchers and PhD-level academics, and a panel of experts in music, AI, and generative technology reviewed it.
Naturally, Investors are still very cautious. Their normal reaction is that no one has done it before, and they want a comparable company that doesn't exist. Every VC wants to see a path to profitability, and the African market adds uncertainty. Ogbeide is very keen to resist the edtech label and sees Bina as a human assistant. "My call is creativity first, then learning," he says. For now, it's a parent tool, with schools to come. A Primary 6 class in Nigeria could use one code, trained on their own social studies curriculum. The infrastructure is very expensive (GPUs and training cost real money), so he is looking at partnerships.
Ogbeide has no finished playbook, and he has no problem admitting it. He does have one rule he wants to pass on: test as you build, and one measure of success. "It's one thing to have an idea," he says. "It's one thing to bring your working software back to the school and have the kids actually use it."
© 2026 Afreoriginals.