Soynade Research is dedicated to democratizing technology for West African languages. We build open-source AI models specifically designed for oral languages. "Soynade" means "to see far" in Fula. To see further than the horizon. We envision a digital future where every African language is a first-class citizen of the internet, supported by sovereign, offline-capable, and culturally grounded AI technologies.
We're on a mission to bridge Africa's digital divide by providing AI tools in native West African languages. By prioritizing speech for oral communities and offline functionality for low-connectivity environments, we enable voice-based services, from mobile money to agricultural support, shaped by users' social and cultural contexts. We're building technology that serves Africa, by Africans, for Africans.
Africa's native languages remain critically underserved by AI, creating barriers in education, healthcare, and information access. In West Africa, where many languages are primarily oral, this exclusion deepens existing disparities. Current AI models fail to capture the linguistic and cultural nuances of African languages, while limited connectivity makes most AI solutions inaccessible. The need for offline-capable, culturally relevant AI has never been more urgent.
We build open-source AI solutions combining advanced LLM architectures optimized for low-resource languages, culturally relevant datasets, speech integration for oral traditions, and offline functionality.
Why Speech-First?
Recognizing that most speakers cannot write their languages, we prioritize speech capabilities throughout our technology stack. This enables voice-based financial transactions, agricultural diagnostics through image analysis with explanations in native languages, and accessible information services.
Our Technologies:
Oolel: Wolof Text LLM
A small, frugal Wolof language model that runs completely offline on standard smartphones, designed for question-answering and education, the first open LLM for Wolof.
Wolof Speech LLM
A multimodal version of Oolel that takes speech as direct input. Beyond Oolel's text capabilities, this model handles speech-to-text transcription and translation with high accuracy on local accents, generates natural-sounding text-to-speech, and enables voice cloning. The key innovation: it understands how West Africans actually speak with mixed languages, regional accents, and the fluid way people communicate in real conversations.
Speech-Text Embedding Models
These create a unified vector space where speech and text can be compared directly, unlocking powerful use cases: RAG applications where users can query knowledge bases using either voice or text, multimodal search systems, and semantic understanding across modalities. For example, a farmer can ask agricultural questions in Wolof by voice and get accurate answers from a database.
Released Models:
We've already released the best-performing models for Fula (translation across English, French, Arabic, plus speech recognition) and Wolof (the first open LLM for the language).
Our business model is built on a powerful flywheel: open source drives adoption, while enterprise solutions drive revenue. Our open-source models act as our marketing engine, building a massive community of developers and creating demand. We then monetize by providing value-added services on top of that foundation through three distinct tiers: The Managed API Platform: This is our "Instant Access" model. We remove the barrier to entry for developers and tech-enabled businesses. They get pay-as-you-go access to our unified multimodal capabilities, speech and text, without any infrastructure headaches. Custom & Private Enterprise Models: This is our "Sovereignty & Security" model, designed for regulated sectors like banking and government. We create custom versions trained on their private data and deployed in their own secure environments, even offline, ensuring total data privacy and compliance. Specialized Industry Solutions: This is our "High-Value Vertical" model. Instead of just raw technology, we sell end-to-end business outcomes. This includes automated dubbing workflows for the movie industry and voice-based transaction engines for financial services, solving specific, high-value B2B problems.
Want to raise from VCs
Global