MongoDB adds vector search to self-managed editions to power generative AI apps

MongoDB has extended vector search and other capabilities to its self-managed database offerings — Enterprise Server and Community edition — to help developers build generative AI and agentic applications.

The same capabilities were added to its managed database offering — Atlas — in June 2023 with the same intent.

Both the Enterprise Server and Community Edition require users to manage their own deployments. The key difference: Enterprise needs a paid license, while Community Edition is free and open source.

The addition of vector search and new capabilities, according to the NoSQL document database provider, will help enterprises using these versions bypass the challenge of relying on external search engines or vector databases to build AI-driven applications, just like in Atlas.

The reliance on a fragmented, heterogeneous stack also introduced operational overhead along with complex extract, transform, and load (ETL) pipelines that were prone to synchronization errors, leading to higher costs, MongoDB said.

Typically, developers need vector search for building AI-driven applications as it is faster and provides more relevant results to a query presented.

Vector search uses mathematical representations to find and retrieve data based on contextual similarity, rather than searching for exact matches.

This ability to conduct similarity searches can also be used by developers to build retrieval-augmented generation (RAG) systems that enhance the reliability of large language models or agents based on them by grounding their outputs in verified enterprise data and content.

The addition of vector search to the self-managed offerings, according to MongoDB, also opens the door for enterprises to pair MongoDB’s native vector search with popular open-source frameworks like LangChain and LlamaIndex, making it easier to build (RAG) applications on self-managed infrastructure.

Industry analysts see this move as more than just a technical upgrade. Jason Andersen, principal analyst at Moor Insights & Strategy, views the new capabilities as part of MongoDB’s broader strategy to attract a wider customer base.

“Enterprise Server drives significant revenue for MongoDB,” Andersen said.

The NoSQL document database provider has been aggressively strategizing to garner more customers as all database providers continue to evolve vector search and other capabilities for building AI-based applications in their offerings.

While at one end, traditional database players, such as MongoDB, Google, etc., have added vector capabilities, specialty vector databases are adding additional features to make their products more easily consumable by non-experts.

When asked about the delay in releasing vector search to the self-managed offerings, Andersen pointed out that it could have been a business decision for the company to prioritize Atlas as it is their flagship offering.

Vector search and other capabilities added to the managed offerings are currently in public preview.