For most, the arrival of capable search engines is the singular event that brought the internet mainstream. Decades on, we’re at an inflection point where artificial intelligence is forcing a rethink of the search concept.

Neeva Inc. succeeded in taking AI and large language models and porting them to consumer search. Following its acquisition by Snowflake Inc., the technology will now exist as a secure, governed and simplistic way for organizations to train LLMs at scale.

“It is really this combination of generative AI, large language models and search. That was the magic unlock for us,” said Sridhar Ramaswamy (pictured), senior vice president at Snowflake and co-founder of Neeva. “That was the product that we released early this year. Fast forward to today, we are very excited about bringing retrieval, bringing search, bringing the power of language models to all of the things that happen within Snowflake.”

Ramaswamy spoke with industry analysts Dave Vellante and Lisa Martin at Snowflake Summit, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the tangible results Snowflake unlocks through Neeva’s AI search technology. (* Disclosure below.)

A revamping of the in-house approach

Just as individuals interact with the internet through search, businesses use it to find and collate mission-critical data. Neeva makes search more intelligent at scale by harnessing natural language, thus expediting the discovery of data points, assets and insights and maximizing their value. A granular ability to discern and filter useful data based on the query, therefore, is Neeva’s cornerstone, according to Ramaswamy.

“But as I said, generative AI was a big unlock for us, because we said, ‘we can get away from giving you a set of links where you have to do the work to go figure out maybe it’s the 40th paragraph that has the answer to the question that you’re looking for,'” he explained. “What it let us do was figure out the right components of the page that supply the answer for you and give you this really pithy, understandable, but believable summary of what you needed to see.”

Targeted at Snowflake’s developers, customers and product partners, the functionality is set to elevate conversations and search within the platform. So instead of having to perform complex processes to retrieve already-stored data, users benefit by simply asking questions within the platform, Ramaswamy added.

“There’s a wealth of information that is sitting out there, but all of us know that this act of actually getting insights out of that data takes work,” he noted. “You have to write the right kind of SQL queries or maybe you will write it in Python with something such as PySpark. But you have to do the work to make those things happen. What we are excited by is combining our expertise in retrieval, in search with the query engine, with other tools such as visualization that’s already there in Snowflake to create, for example, conversational experiences.”

The ability to perform at scale and extract maximum output out of minimal resources is Neeva’s main differentiator, according to Ramaswamy.

“Even within Neeva, even with the 50% team on a very small budget, we ran a crawl that every day fetched hundreds of millions of pages,” he said. “Our index was 5, 6 petabytes and then we serve 6, 7 billion pages on machines that cost $150,000 a year — so, doing a lot of stuff with very few dollars.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Snowflake Summit:

(* Disclosure: TheCUBE is a paid media partner for Snowflake Summit. Neither Snowflake, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE