Generative AI solutions with Google and Elasticsearch – SiliconANGLE

When Google LLC announced new generative AI features for its cloud platform in April, the company provided users with an ability to use technology from Elasticsearch Inc. to leverage cross-network data for model training.

The announcement highlighted how Google Cloud has been working with its partners to offer a one-stop shop for enterprise generative artificial intelligence development.

“Every customer has a … data problem,” said Ken Exner (pictured, right), chief product officer at Elastic. The combination of Google’s data processing and storage with Elastic’s search, “analytics and AI capabilities create a powerful way to sift through all that data to get to insights and to get to things that they can [take] action on,” Exner further explained.

Exner spoke with theCUBE Research’s John Furrier for the Google Cloud Marketplace Marvels interview series, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. He was joined by Stephen Orban (left), vice president of migrations, ISVs and marketplace at Google Cloud, and they discussed how the partnership is providing customers with innovative ways to deploy and manage generative AI. (* Disclosure below.)

Building advantage through use of private data for generative AI

Elastic’s solutions bring together intelligent content search, IT observability and security analytics to help customers access critical data for enterprise applications. With increased adoption of generative AI, businesses are looking for ways to gain a competitive edge through the use of private data.

“When you talk about data and what the impact of that is in customers’ generative AI pursuits, the models that we make available are great but they’re currently trained on publicly available data,” Orban noted. “Customers who are doing unique things with their generative AI capabilities … they need to be either tuned or augmented using, let’s say, retrieval augmented generation with data that they own in their own private corpuses. Oftentimes, customers are already using Elastic as the storage layer for that corpus of data.”

A desire to access public and private data stores represents a change in how customers have viewed data in the past, according to Exner.

“One of the things we’ve been seeing in this industry for a while is how do you deal with all this data?” he said. “Typically, people have been trying to reduce the amount of data that they have. What’s different these days is that it’s OK to have all that data if you can search through it and parse it and then pass it to an LLM and have an LLM help you get answers out of it. Actually, the more data, the better.”

Unique capabilities drive expanding use cases

Having more data has placed greater emphasis on automation and speed. Companies are looking to implement generative AI for a wide range of uses which, in turn, is driving Google Cloud’s partnership with Elastic.

“Generative AI is going to make way for so many use cases we haven’t even thought of yet, but also automate a lot of existing things today, whether it be customer service or summarizing documents or generating code or images or marketing campaigns,” Orban said. “The unique set of capabilities that Elastic has built with us over the years … really position us well to help customers, not just identify those use cases but implement them very quickly.”

The implementation of use cases requires an ability to leverage inferencing tools provided by Google Cloud on the Elastic platform. These include Vertex AI, which is used to serve large language models, and Gemini Pro, part of Google Cloud’s family of generative AI offerings.

“In the inference part of this, we not only support our own inference model, we support Vertex AI embeddings creation services so that you can do that directly from the [open] inference API in Elasticsearch, and we support passing context to an LLM at the end of that workflow, in this case Gemini Pro,” Exner said. “You can use Elastic … with the best generative AI at Google to create a complete workflow for grounding and building an LLM-based application on top of Google and Elastic.”

This combination has been helpful for customers seeking support in areas such as security, according to Exner.

“You want to have anomaly detection that runs to find things that you can’t necessarily predefine as an alert,” he explained. “The combination of predefined alerts and anomaly detection based on generative AI and machine learning models will allow the operations analysts to get the best of both worlds. Because of the integrations that we have with Gemini and the LLMs we can start automating the remediation workflow.”

Google Cloud is working with Elastic to build new features, including the use of Gemini Code Assist, a service inside of Google Cloud Console where customers can ask questions and receive natural language responses.

“There’s more that we can do specifically to make sure that when a customer is starting to build or train a model, they could discover an Elastic data source that they’re already running in GCP without necessarily having to leave the experience that they’re in,” Orban said. “We’ve started that integration already by Elastic working with us and training our Gemini Code Assist with Elastic specific data. There’s a lot more that we can and will be doing as the future unfolds.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE Research’s coverage of the Google Cloud Marketplace Marvels interview series:

(* Disclosure: TheCUBE is a paid media partner for the Google Cloud Marketplace Marvels special interview series. Neither Google LLC, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE