After Snowflake acquisition, former Neeva CEO says ‘window is shutting’ for AI search disruption

In a wide-ranging interview with VentureBeat, former Neeva CEO Sridhar Ramaswamy — who became SVP of data cloud company Snowflake after the company acquired the ad-free, privacy-focused, AI-powered alternative to Google Search — said the “the window is shutting” for AI chatbots to disrupt Google’s consumer search dominance.

“It seems like there was a tiny sliver to truly disrupt consumer search,” he told VentureBeat in an in-person interview at Snowflake’s Manhattan office before jetting off to meet Snowflake customers in New Zealand. “What I mean by that is in an ideal world, you would have ChatGPT and Bing working together to create a believable ChatGPT. They’re not. On the other hand, Sidney, which is the official chatbot from Bing, is getting no market share and is just really, really slow.”

Meanwhile, Google has been slowly but surely working on their own generative search experience called Bard, which which Ramaswamy says is “a little slow but is actually getting better.” So while consumer search powered by language models looked very promising early this year, he explained, and he is “proud of what we did at Neeva,” the window of true disruption to Google search is closing.

Moving into enterprise AI

Neeva decided to move to the enterprise search space — through the Snowflake acquisition — because “it’s a far more predictable way to create value,” said Ramaswamy. At Neeva, he explained, growth was driven more “by the whims of my friends at Apple, on things like will they add another search provider to Safari, than by the quality of our product. So you did feel like the enterprise space was more rational.”

Ramaswamy, who before co-founding Neeva led Google’s $115 billion advertising division, said the first three months at Snowflake has been all about understanding the company’s product roadmap and what the Neeva team will build to help Snowflake clients with broad-based search access to structured and unstructured data.

“One smart thing we did is start building prototypes on top of Snowflake even before the acquisition,” he said. “My cofounder and I forced the team to start thinking about what would we build? So we got a little bit of a running start.” Now, he added, he feels ready to tackle deep customer conversations that helps clients move on their Snowflake journey.

“Snowflake has spent the better part of 10-plus years building up amazing relationships with customers, they trust us with their data, and in turn, we take their data very, very seriously and make sure it’s safe and secure for them,” he said. A lot of enterprises, he added, turn to Snowflake to ask how they should use the power of AI in the experiences they create, both internally and for customers: “So this feels like a really good place to be.”

Of course, enterprise search and knowledge discovery is becoming a highly-competitive space, with companies like Glean in the mix. But Ramaswamy emphasizes that Glean is about internal document search, a very different offering than what Neeva’s technology provides to Snowflake — that is, access to both structured and unstructured data for whatever requirements and enterprise company needs.

AI success remains ‘anyone’s game’

Ramaswamy said that developments in large language models continue to be interesting: “They are the new iPhone screen, the new keyboard and mouse,” he said, in terms of breakthroughs. That said, it’s unclear who will reap the benefits of AI.

“I don’t think anybody knows,” he said. “It’s also changed — in 2007, no one could have predicted that BlackBerry would be gone, that Nokia would be gone, and… a shitty operating system from Google would dominate the world.”

Platforms, in particular, are hard to predict as far as who reaps the value, he added. “Windows famously made enormous amounts of money for Microsoft. For Google, it was mostly a defensive play to keep their search business alive and thriving on mobile. I would say the platform strategy for Apple iOS is incredibly well-executed.”

When it comes to AI models, “I don’t really think anyone knows,” he emphasized. “In the space of the last six months, we have gone from, oh crap, three companies are going to sweep the table, like they did with the cloud — to wow, it’s really anyone’s game.”