Document databases, AI embeddings and retrieval
MongoDB / Voyage AI
MongoDB provides a document database and data platform for developers building applications, while Voyage AI supplies embedding and reranking models for retrieving relevant information from documents, code, images and enterprise knowledge bases. MongoDB Atlas combines managed database services with search, vector search and native access to Voyage models. Enterprise Advanced offers self-managed deployment in cloud, on-premises or hybrid environments.
Dwight Merriman, Eliot Horowitz and Kevin Ryan founded MongoDB’s predecessor, 10gen, in 2007; Chirantan “CJ” Desai is its current president and CEO. Voyage’s co-founders include Tengyu Ma and Hong Liu, who previously led its research. Its engineering work includes a shared embedding space across the Voyage 4 family: developers can embed documents with a larger model and queries with a smaller one, then change query models without re-embedding documents. The family also includes a mixture-of-experts model and open weights for local development.
MongoDB completed its acquisition of Voyage AI in February 2025, with purchase consideration valued at $160.9 million. Voyage models are available through both the Voyage API and Atlas. MongoDB reported more than 63,900 Atlas customers as of January 31, 2026, excluding free users and counting affiliated entities together.
1 talk
Newest first
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Messages from the stage
What stronger retrieval models could replace
Ma predicts that increasingly capable retrieval models will absorb more of the hand-engineered techniques used in RAG. His technical survey includes Matryoshka learning, quantization, contextual retrieval, and domain-specific code embeddings.
Affiliations reflect each recorded session, not necessarily current employment.