Today, we are excited to announce that we’ve joined MongoDB!

When we founded Voyage AI in 2023, our mission was to help developers build the best retrieval systems to power intelligent AI applications. AI is already transforming industries at a rapid pace, but for it to reach its full potential, its outputs must be grounded. Embedding and reranking models—though often overlooked—play a crucial role in achieving this reliability, making them an essential part of the broader AI stack. 

I am proud of the progress we were able to make in a very short period of time. In just over a year and a half, we built a world-class research team, released high-quality embedding and reranking models that consistently ranked at the top of public benchmarks, and collaborated with some of the most exciting AI innovators in Silicon Valley, like Harvey, Replit, and more. Our latest models—including Voyage AI’s general-purpose, multilingual embedding model (voyage-3-large) and our code retrieval model (voyage-code-3)—attain state-of-the-art retrieval accuracy and dramatically reduce vector storage costs.

Despite the exciting journey we were on, we recognize the huge opportunity that joining MongoDB presents. 

Developers today face major challenges when building AI applications—they must stitch together multiple components, from vector databases to operational databases, rerankers, and embedding models, just to create retrieval-augmented generation (RAG) and agent systems. This fragmented approach is complex, costly, and inefficient. We believe that a more unified stack—where core AI components live directly in the database—will lead to a smoother, more productive developer experience while also improving the performance and accuracy of AI-powered applications. By combining MongoDB’s robust infrastructure with Voyage’s cutting-edge retrieval models, we are creating a stronger, more cohesive solution for AI developers.

Furthermore, from the perspective of large enterprises, modern AI is fundamentally about extracting meaningful semantic insights from large-scale unstructured data. MongoDB’s document model is uniquely suited for this, as it seamlessly represents structured, semi-structured, and unstructured data. We firmly believe that the next generation of AI applications will be built on MongoDB, making it the ideal foundation for AI-driven systems. To that end, joining MongoDB allows us to offer our models to a much wider audience, unlocking the value of the world’s unstructured data in ways that were not possible before. Together, we empower organizations to build scalable AI applications with greater accuracy and reliability. 

Finally, we were incredibly impressed and inspired by the MongoDB leadership. After meeting Dev Ittycheria (MongoDB’s CEO) and Sahir Azam (Chief Product Officer), we were inspired by their vision of bringing AI-powered retrieval directly into the database by combining the strengths of MongoDB and Voyage AI. Their conviction, efficiency, and innovative perspective are truly defining what the future of AI-powered databases should look like. Dev and Sahir are also true believers in the importance and potential of AI, something we—a group of academics and AI researchers who live and breathe AI—were looking for in any potential partner.

We’re excited about what the future holds and can’t wait for our customers to benefit from this new partnership. For more on why MongoDB acquired Voyage AI, and how our models will be integrated into MongoDB, see Dev’s blog post.

Tengyu Ma