Shay Banon

And we're saying internally at Elastic that we want to make Apache, Lucene and Elasticsearch the best vector database in the world. And then on top of it, the best hybrid search, because it's like it's not only vector search, right? It's like there's a lot of layers on top of it to make a great search system that we've been building for years.

And we're saying internally at Elastic that we want to make Apache, Lucene and Elasticsearch the best vector database in the world. And then on top of it, the best hybrid search, because it's like it's not only vector search, right? It's like there's a lot of layers on top of it to make a great search system that we've been building for years.

And yeah, I mean, like any new type of capability, I think you end up having like new competition or other new vendors or products, open source or not, as you say. I think Elasticsearch is a... Great vector database today. In Apache, we contributed so much work into Apache Lucene.

And yeah, I mean, like any new type of capability, I think you end up having like new competition or other new vendors or products, open source or not, as you say. I think Elasticsearch is a... Great vector database today. In Apache, we contributed so much work into Apache Lucene.

We have some of our best developers working just on Apache Lucene on making it better and then integrating it into Elasticsearch. By the way, by making Apache Lucene better, we're making a lot of other systems better. So it's like, our focus is to put the right technology at the right place and then win based on value.

We have some of our best developers working just on Apache Lucene on making it better and then integrating it into Elasticsearch. By the way, by making Apache Lucene better, we're making a lot of other systems better. So it's like, our focus is to put the right technology at the right place and then win based on value.

And yeah, and so far, by the way, that's the fastest growing market in the world, in our business. So that's definitely the fastest part for us in terms of usage and business. And people are seeing the value. I do think one of the things that I'm trying, working really hard to figure out is one of the things I think Elasticsearch did really well is that it made search approachable.

And yeah, and so far, by the way, that's the fastest growing market in the world, in our business. So that's definitely the fastest part for us in terms of usage and business. And people are seeing the value. I do think one of the things that I'm trying, working really hard to figure out is one of the things I think Elasticsearch did really well is that it made search approachable.

You could just dump a JSON document into it and start to search it, and you didn't have to understand vector space and ranking and BM25 and inverted indices and FSTs and all of these things. And then we progressively disclosed complexity if you wanted to or you wanted to tweak things, but it just worked.

You could just dump a JSON document into it and start to search it, and you didn't have to understand vector space and ranking and BM25 and inverted indices and FSTs and all of these things. And then we progressively disclosed complexity if you wanted to or you wanted to tweak things, but it just worked.

And we've invested over the last year a lot of effort, not only in making us a great vector database and great in hybrid search and things like that, but also making it as easy as it was to index. We literally took the curl requests that I had in the first blog post when I released Elasticsearch around indexing tweets, and we're literally trying to make it this simple to do semantic text.

And we've invested over the last year a lot of effort, not only in making us a great vector database and great in hybrid search and things like that, but also making it as easy as it was to index. We literally took the curl requests that I had in the first blog post when I released Elasticsearch around indexing tweets, and we're literally trying to make it this simple to do semantic text.

And I think if we go back to hundreds of thousands, if not millions of developers that know Elasticsearch and know Elasticsearch API, that's a huge opportunity because they don't have to become experts in vector databases and product quantization and HSNW and dimensions and embeddings models and things like that.

And I think if we go back to hundreds of thousands, if not millions of developers that know Elasticsearch and know Elasticsearch API, that's a huge opportunity because they don't have to become experts in vector databases and product quantization and HSNW and dimensions and embeddings models and things like that.

We can really simplify that layer to them and they can use Elasticsearch the way that they love.

We can really simplify that layer to them and they can use Elasticsearch the way that they love.

And to me, like, that's a huge opportunity because if we make every single Elasticsearch developer out there, like someone that uses Elasticsearch, just use semantic search, even without, almost without noticing, because it's so simple, like that would be amazing for them because it's significantly better. And for us as a, as a, as a product in Elasticsearch and as a company.

And to me, like, that's a huge opportunity because if we make every single Elasticsearch developer out there, like someone that uses Elasticsearch, just use semantic search, even without, almost without noticing, because it's so simple, like that would be amazing for them because it's significantly better. And for us as a, as a, as a product in Elasticsearch and as a company.

I think it's here. So we, I don't know, I'll get geeky for saying, but we have like a, when you define something in Elasticsearch, there's mappings for it, which is basically the schema. And then there's like text mapping, which basically does keyword-based like analyzer, BM25 type matching. If you do semantic underscore text and don't change anything else, it just becomes semantic.

I think it's here. So we, I don't know, I'll get geeky for saying, but we have like a, when you define something in Elasticsearch, there's mappings for it, which is basically the schema. And then there's like text mapping, which basically does keyword-based like analyzer, BM25 type matching. If you do semantic underscore text and don't change anything else, it just becomes semantic.