Some of these were around before AI became a buzzword, but these startups are working hard to make sure Ethiopia isn’t left behind on what’s popularly referred to as “the fourth industrial revolution".
. . .
AI is an inescapable reality of the 21st century. Given its ubiquity, Ethiopia has made commendable efforts to adopt AI technologies. In June 2024, Ethiopia officially announced a National Artificial Intelligence policy, which is expected to provide legislative guidance and outline a vision for what AI would look like in Ethiopia. In the meantime, several AI startups have popped up. Granted, some of these were around before AI became a buzzword, but these startups are working hard to make sure Ethiopia isn’t left behind on what’s popularly referred to as “the fourth industrial revolution”.
In this article, we’ve compiled a list of some of the active Ethiopian startups working in the AI sector.
Lesan AI
Based in Germany, Lesan AI was founded in 2019 by Asmelash Teka. The company leverages an NLP (Natural Language Processing) model to provide translation services for Amharic and Tigrigna languages. Dissatisfied with the little attention paid to African languages by translation services such as Google, Asmelash said that he decided to create a customized model that would better perform for Ethiopian languages. While seemingly a small initiative, Lesan is not to be taken lightly. Its translation quality is said to have outperformed that of Google Translate.
Begize AI
Operating since June 2024, Begize AI is a content creation platform that integrates the OpenAI engine to power its system, According to General Manager, Zerubabel Kibebe, their platform combines AI tools from OpenAI, Google, Anthropic, and Cloud APIs, enabling users to leverage a diverse set of AI capabilities for content creation and more. Begize AI provides an all-in-one solution for AI users, helping them leverage multiple models to optimize their performance.
Hasab AI
Though not officially launched yet, HasabAI plans to bridge the technology gap for low-resource languages, focusing on African languages through its Audio Intelligence API. Offering high accuracy, multilingual support, and transcription, HasabAI promotes linguistic diversity while delivering cutting-edge AI solutions.
Leyu AI
High-quality datasets are essential for AI, yet Ethiopian languages remain underrepresented, limiting AI's impact. Leyu (ለዩ), meaning "identify" or "label" in Amharic, tackles this challenge by crowdsourcing datasets for Amharic, Afaan Oromo, Tigrigna, and other low-resource languages. By doing so, Leyu is helping to solve a crucial problem that’s hindering the growth of AI in Ethiopia; the shortage of high-quality and ethically sourced datasets. As of today, Leyu is still in the testing phase and has not officially launched yet.
iCog Labs
Of all the companies on this list, iCog Labs has been in the game the longest. Founded in 2012, the company worked on AI platforms during a time when few people heard of the word “AI”. Known best for their contribution to the building of Sophia the Robot, iCog Labs is also a key contributor to OpenCog; an open-source AI framework. notable for its ambition to contribute to the development of AGI (Artificial General Intelligence).
Zemenu
Zemenu is a startup that’s been around since 2019, but its foray into AI is rather recent. Not too long ago, Zemenu revealed its latest product Lucy; a chatbot and voice assistant that caters to Ethiopian languages. According to the announcement post, Lucy is capable of speaking Amharic, Afaan Oromo, and Tigrinya, as well as English and over 100 international languages. While not officially launched yet, the developers say that Lucy will bridge language barriers, enhance accessibility, and drive digital transformation.
Addis AI
Developed by AiQEM Techs, Addis AI is an Amharic voice assistant, one that can listen and respond in Amharic. Whether you prefer speaking or texting, it understands you either way. And, it’s available on Google Play. According to AiQEM Tech, the chatbot is “designed to deliver accurate, context-aware, and culturally sensitive responses.”
Admas AI
Initiated by the Japan International Cooperation Agency (JICA) and developed by NUNA Ethiopia, with VP Trading driving the tech, Admas AI is a telegram-based chatbot that aims to boost entrepreneurship. It claims to do so by aiding startups and businesses with Concept Sheets, Lean Canvases, and Business Plans.
Abyssinica GPT
Developed by Ethiocloud, Abyssinica is an advanced AI model for Amharic and other African languages. Enabling natural, human-like conversations, a valuable tool for information across customer service, education, and more.
EthioRobot AI
Much like the previous three, EthioRobot AI is a chatbot that has Amharic language support and is customized “ specifically for Ethiopians”. The platform is currently paywalled and requires a subscription to access it.
While these companies are doing impressive work, it’s unclear whether they are developing custom AI models in-house or primarily relying on external models and building interfaces to handle API responses. In simpler terms, are these companies creating their own AI models, or are they using AI models developed elsewhere and tailoring them to their needs?
There appears to be a lack of transparency (with exceptions, of course) in this area, but given the high cost of training and deploying an AI model, it’s likely that they are using pre-existing models and tailoring them to their needs. Even Deepseek, the cheapest trained AI to date, has cost 6 million dollars, which is a kind of money that is not easily available in Ethiopia. So, most of these startups are likely to adopt pre-trained models and fine-tune them with custom datasets (i.e. retrain them to fit a specific purpose). There are exceptions to this. Lesan AI, for example, has developed its own machine translation model.
It’s All About Data
Another issue worth mentioning, one that plagues the entire AI industry, is the sourcing of data. In Ethiopia, where copyright laws are already lax, this becomes an even greater challenge. The lack of robust intellectual property protections makes it difficult to ensure that data is ethically sourced, leading to potential issues around data ownership, quality, and privacy. This creates obstacles for AI companies in building reliable, high-quality datasets while maintaining legal and ethical standards. Startups like Leyu are trying to resolve this by providing ethically sourced data sets.
The Bottom-line
Ethiopian AI startups are building a solid foundation for Ethiopia’s future. While dealing with inadequate resources and a not-too-friendly business climate, these startups are creating innovative solutions that’ll help Ethiopia keep pace (or at least not fall too far behind) in the rapid advance of AI.
