Vector Search: Powering the Next Generation of Applications

While Vector Databases have been around for some time, the advent of the transformer architecture has led to the supercharging of semantic search with vectors. With MongoDB Atlas’s new Vector Search offering, customers can take advantage of this transformative technology on top of their application data.
In this talk, we will focus on core concepts around Vectors, embedding your data, and the range of use cases we see our customers exploring with Vector Search. We’ll then go through a demo where we show what the experience would be when embedding/vectorizing a document, inserting it into the cluster and finally querying that data to find semantically similar data to our questions. Lastly, we’ll talk about some of the new exciting things we’re exploring during our Public Preview. Don’t miss out on this opportunity to learn about the next revolution in building applications.
Learn more → trymongodb.com/44dMMnG
Blog post → trymongodb.com/46g7OUC
View All Sessions → trymongodb.com/mongodblocalnyc23
Subscribe to our channel → mdb.link/subscribe
#MongoDBlocalNYC2023

[@hanslanger4399]

One of the best presentation I have ever seen about an overview of embeddings/ vector data, Thank you for sharing. Ben Flast, you are brilliant, great job!

[@greendsnow]

is there a soccer match going on in the meanwhile?

[@bellahasguns]

Women’s World Cup

[@rajithkumar3424]

Nice presentation . Finally Mongo into Vector Search . way to go

[@NeverReply]

Such a nice feature explaning for a late bed time. But hands down, that was really inspiring!

[@MongoDB]

Thank you for the kind words!

[@darkstudio3170]

Great presentation. One question , how this gonna work in distributed environent ? for suppose a new querry , the nearest neigbours may be present in different nodes / partitions.

[@siddheshshirawale4115]

very informative presentation. How can I implement vector search on already inserted documents ?

[@ajithkumar0]

This is nice. How can we run vector search with a filter for geo location. Its asking for two indexes in the same pipeline, knnVector and geo type indexes - which is not possible as of now?

[@jorgegimenezperez9398]

Damn people in the background are very hyped about vector search!

[@thehappycookiehour]

What about the costs?

[@robertcormia7970]

This was outstanding! I'm just a neophyte, but thinkng about an application involving vector embeddings of complex data, electron spectroscopy, probably a very high dimensional vector.

[@MarcSalvat89]

Thanks for the presentation! It's a very nice feature, but will you release it outside of Atlas for on-premises systems?

[@MongoDB]

For now the feature is only available for MongoDB Atlas

[@Tritoon710]

Would love to see MongoDB as embedded database for desktop application as well.

[@justdoeverything8883]

It is with MongoDB Realm! I just found out about it today actually.

[@adrianthomas5934]

Nice presentation. Are vectors only field level or could the embeddings be for all the fields in a document? Also will standalone servers support this in the near future not just Atlas based? tia

[@pedrorabbi]

@MongoDB Are there any additional costs for using that? Aside for the inherent costs of a little more storage and maybe a little more processing

[@harrykekgmail]

great presentation. thank you. MongoDB, way to go!

[@MongoDB]

We're glad you enjoyed it! 💚