Ideas & vision
When ads speak Africa's two thousand languages
Chrisphine Otieno · 19 Oct 2026 · 2 min read
Africa is home to more than two thousand languages, by most linguists' counts. Kenya alone has dozens. Yet look at the adverts on Kenyan screens and you will see mostly English, a good deal of Kiswahili, and occasionally a vernacular radio spot. Across the continent the pattern repeats: a few dominant languages carry almost all advertising, while the languages that millions speak at home appear rarely.
The reason has been cost. Every new language meant new scripts, new voice actors, new edits. For most businesses, one or two languages were all they could afford.
AI changes the cost
When an AI tool can produce a version of an ad in another language in minutes, the economics of multilingual advertising change. Today, Tari Studio lets businesses set the language of every spoken line and on-screen word to Kiswahili, English as Kenyans speak it, the Kiswahili-English mix, English (international), French or Chinese (Mandarin). Ads in your customers' languages.
That is a start. The bigger opportunity lies in the many languages not yet well served by AI. Across the continent, efforts are under way to change that: speech models, datasets and language-identification tools for dozens of African languages. The year of African-language AI.
What becomes possible
Imagine, as a vision rather than a promise:
- A health campaign in Turkana, Luo, Kalenjin and Kikuyu, made in an afternoon
- A seed company advertising in each region's language during planting season
- A bank explaining a savings product in Luganda, Hausa or isiZulu
- A local business speaking to customers in the language of their grandmothers
Why it matters beyond marketing
Language is identity. When a language appears in modern media, its speakers see it as part of the future, not only the past. Young people are more likely to keep speaking a language they hear in music, film and, yes, advertising.
The care it requires
Bringing more languages into AI advertising must be done carefully:
- Accuracy: bad translations are worse than none. Native speakers must check every word.
- Respect: languages are not costumes. Do not use one to mock or stereotype its speakers.
- Data ethics: building AI for a language needs recordings and text from its speakers, with consent and fair benefit.
- Dialects and registers: a language spoken across a region varies. Get the right form for your audience.
- Community involvement: speakers should shape how their language is used.
A view
We believe the future of advertising in Africa is multilingual, and that AI can help get there faster. But languages belong to their speakers. The tools should serve them, not flatten them. Every new language we and others add should be added with the people who speak it.