MongoDB launches Voyage AI, search tools aimed at on-prem, private cloud deployments

MongoDB added new tools to Voyage AI and its search to improve accuracy for on-premises and private cloud deployments that enable AI applications to run anywhere.

The company, which announced efforts to improve AI accuracy at MongoDB.local Bengaluru, is addressing the reality that a big chunk of AI workloads will run outside of the cloud due to cost, latency and keeping data in-house. MongoDB's play is to improve accuracy of AI apps.

• MongoDB raises outlook after strong Q1 results
• MongoDB adds automated Voyage embeddings in Atlas Vector Search
• MongoDB integrates Voyage 4 models across its platform

MongoDB said its Voyage Context 4, Hybrid Search and Native Reranking work together to improve accuracy. The general availability of Search and Vector Search in MongoDB Enterprise Advanced and Community Edition will add the same retrieval tools in its Atlas platform for on-premises, private cloud and local environments.

According to the company, enterprise AI projects hit the wall due to memory, retrieval, accuracy and compliance. Here's a look at what MongoDB announced as additions to its database:

• Native Reranking in MongoDB Atlas, which is in public preview, runs Voyage AI and features up to a 30% boost in retrieval quality inside the database.
• Voyage Context 4, generally available, is a new embedding model for long documents. It processes long documents in full context instead of isolated chunks. Voyage Context 4 drops into retrieval augmented generation pipelines without re-architecting.
• Hybrid Search in MongoDB is generally available and combines full-text and vector search in one query inside the operational database. Embeddings stay up to date.
• MongoDB Search and Vector Search are generally available as an add-on to MongoDB Enterprise Advanced as well as MongoDB Community Edition.
• MongoDB Atlas Stream Processing with Apache Iceberg Support is generally available and can enable any Atlas collection to be continually synchronized with AWS. The company also said Gen2 MongoDB Atlas M30+ Dedicated Clusters on AWS is generally available.

The company cited Infosys and Emergent Labs as key customers of the new capabilities. MongoDB also announced plans to upskill 2 million Indian developers by 2030, expand its MongoDB for Academia program and launch a startup challenge.