MongoDB Vector Search : Video

MongoDB Vector Search : Video

Vector Search: Powering the next generation of applications.

Last week, Pureinsights attended MongoDB.local in NYC, a user conference for MongoDB’s Developer Data Platform.  It was exciting to hear about new search-related product enhancements to Atlas Search, and other highlights like the Keynote, which we summarized in our blog.  For us, the most exciting MongoDB announcement was a preview of Vector Search.

MongoDB’s integration of vector database capabilities into the platform to support vector search and semantic search use cases opens up a whole new generation of applications.  Ben Flast, MongoDB Lead Product Manager, did an excellent presentation on the topic, so we are sharing it directly below.

In his presentation, Ben reviews core concepts around Vectors, embedding your data, and the range of use cases customers are exploring with Vector Search.  He goes through a demo of embedding/vectorizing a document, inserting it into the cluster and finally querying that data to find semantically similar data to our questions.

Vector search works better than traditional keyword search when queries may be full natural language questions or a “vague query” not fully expressed in exact terms, like “that movie where Tom Hanks is stuck in an airport” (The Terminal).

Ben Flast explains MongoDB Vector Search

Vectors work in this case because they are better able to model abstract concepts and queries. Vectors are mathematical representations of unstructured data, such as text, images, videos, and audio files, which enable similarity and relevance-based queries without relying on keyword matching. With Atlas Vector Search, users can store, index, and query vectors alongside operational and transactional data in documents, eliminating the need for separate database systems. This addition enables teams to deliver more context-aware results and augment applications built on Large Language Models (LLMs) with proprietary data, improving accuracy and performance.

This really does open up a whole new class of search-powered applications and we’re just scratching the surface of what’s possible.

If you have any questions about vector search applications or MongoDB Atlas Search, please CONTACT US.

Pureinsights Related Resources

MongoDB Related Resources


Stay up to date with our latest insights!