Elastic®, the company behind Elasticsearch®, today announced the release of the Elasticsearch Relevance Engine™ (ESRE™), developed with vector search and the Transformer model built-in. The technology was designed to allow companies to benefit from the power of artificial intelligence at the service of their proprietary data. Enterprises benefit from true technological advances because thanks to ESRE, they can create secure deployments to take advantage of all their proprietary data, structured and unstructured. Thus they can optimize their infrastructure and make the best use of the skills of their employees.
“Generative AI is a critical moment in the evolution of technology, and companies that take the lead in harnessing it appropriately will be the leaders of tomorrow,” said Ash Kulkarni, CEO of Elastic.
“Elastic’s expertise in search is evident in its approach to integrating generative AI into its solutions,” said Julia Lewison, president of the Microsoft Developer Division.
Elastic has made significant investments in developing AI and machine learning capabilities and features to democratize them for developers. Thanks to unified APIs for Vector, Hybrid and BM25f research, but also thanks to a new transformation model that is small enough to install in a laptop’s memory, companies and teams are now able to optimize their infrastructure and their talent in a more efficient way.
Using tools like ESRE allows companies to take advantage of all of their structured or unstructured data. This allows them to design custom applications using AI, without worrying about the size and cost of operating large language models. Companies can use their own transformation model and integrate it with third-party models to build secure solutions, and take advantage of all the benefits generative artificial intelligence has to offer to process their own business data. With ESRE, thousands of enterprises as well as the vast community of Internet users who have invested in Elastic solutions can now benefit from AI, without resorting to many additional resources.
“Enterprises are excited about the impact of generic AI in their applications and workflows. However, the rapid pace of innovation in this area seems inaccessible to them,” said James Governor, co-founder of RedMonk.
The result of more than two years of research and development, ESRE is already used by many Elastic customers to improve AI-based applications. Additionally, Relativity, a global technology solutions provider for the legal industry and an OEM partner of Elastic, is currently testing ESRE with Azure OpenAI Services to understand and demonstrate how this technology can improve the relevance of results from their eDiscovery solution, RelativityOne.
“It is critical to ensure that our customers and partners have industry-leading search capabilities to help them organize their data, deliver verified results, and act on them. I am excited to be able to bring such benefits to our customers through the investment that allows us to leverage Elasticsearch in Relativity One,” said Chris Brown, Chief Product Officer, Relativity.
The Elasticsearch Relevance Engine (ESRE) includes:
advanced relevance ranking features, including BM25f, a fundamental element of hybrid search;
a vector database for storing and searching data in high-dimensional content;
a new proprietary transformation model that includes out-of-the-box semantic search;
the ability to use your own transformer model;
Integration with third-party transformation models such as OpenAI GPT 3/4 via APIs.
Other innovations are under development.
(tagstotranslate)elastic