New unified data intelligence layer helps developers cut hallucinations, simplify architectures and scale AI workloads without duplicating data
MongoDB has announced a major expansion of its AI platform at MongoDB.local San Francisco, unveiling industry-first integration between its core database and Voyage AI’s latest embedding and reranking models. The update introduces a unified data intelligence layer designed to help developers build accurate, reliable production AI applications without moving or duplicating data — a key challenge for fast-scaling AI teams.
“Developers want fewer moving parts and clearer paths from prototype to production. MongoDB is raising the bar for real-world, mission‑critical AI applications.” — Fred Roma, SVP Product & Engineering, MongoDB
The company launched five new Voyage AI embedding models, automated vector embedding capabilities for MongoDB Community, embedding and reranking APIs within Atlas, and an AI-powered operations assistant across MongoDB Compass and Atlas Data Explorer. Together, these additions make MongoDB one of the most comprehensive AI-ready data platforms used by more than 60,000 global customers.
Fred Roma, Senior Vice President of Product and Engineering, said organisations are struggling not with experimentation, but with operationalising AI at scale. “Developers want fewer moving parts and clearer paths from prototype to production,” he noted. “With today’s launches, MongoDB is helping teams reduce complexity and build AI that performs reliably in mission-critical environments.”
“MongoDB allows us to focus on what matters most — our customers and our business — instead of stitching complex AI pipelines.” — Rotem Weiss, CEO, Tavily
The new Voyage 4 series delivers state-of-the-art retrieval accuracy and includes high-performance models — voyage-4, voyage-4-large, voyage-4-lite, and the open-weights voyage-4-nano — outperforming several leading alternatives in independent retrieval benchmarks. MongoDB is also expanding multimodal retrieval with voyage‑multimodal‑3.5, enabling unified vector processing across text, images and video.
A key highlight is Automated Embedding for MongoDB Vector Search, which generates embeddings natively inside the database whenever data is inserted, updated or queried, removing the need for separate pipelines and sync jobs.
Early adopters, including TinyFish and Tavily, showcase the impact. “We were looking for extremely accurate models, and Voyage AI delivered accuracy at scale,” said Sudheesh Nair, CEO of TinyFish.
By unifying operational data, retrieval pipelines and model APIs in a single system, MongoDB aims to help enterprises build faster, reduce latency and ensure AI applications behave predictably in production environments.
