Gunning for Google with Perplexity CEO Aravind Srinivas

**Nathan Labenz** (0:00)
Turpentine is a network of podcasts, newsletters, and more covering tech, business, and culture, all from the perspective of industry insiders and experts.
We're the network behind the show you're listening to right now.
At Turpentine, we're building the first media outlet for tech people by tech people. We have a slate of hit shows across a range of topics and industries, from AI with Cognitive Revolution, to Econ 102 with Noah Smith. Our other shows drive the conversation in tech with the most interesting thinkers, founders, and investors like Moment of Zen and my show Upstream. We're looking for industry-leading hosts and shows along with sponsors. If you think that might be you or your company, email me at erikaturpentine.co. That's E-R-I-K at turpentine.co.

**Aravind Srinivas** (0:45)
We are building our own search index and SOAS, OpenAI, SOAS, Anthropic. Everybody's building their index because I think in a world where large language models are a commodity and the training recipe for them or the weights are just running on their open source, the edge goes to the data markets. People who own the best data in the world. What do I mean by that? It's like there are like trillion pages on the web. We can't index all of them. So then you narrow it down. You don't even want 100 billion pages in your index. It's not about the quantity here again, right?
You want the best web pages on the internet. That's probably a billion or 10 billion, I don't know.
But the best ones that really matter to the knowledge worker, to the researcher, to the curious mind, right? If you can capture that distribution really well, there's already a huge moat there. I think there's very few companies that can aim to do this.
It's like a chicken and egg problem. In order to do this, you need to have a product. But in order to have a product, you need to have some kind of index. But somehow, we broke that asymmetry. So we are at a point where we can dream of being important.

**Nathan Labenz** (1:58)
Hello, and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs, and builders working on the frontier of artificial intelligence. Each week, we'll explore their revolutionary ideas, and together, we'll build a picture of how AI technology will transform work, life, and society in the coming years. I'm Nathan Labenz, joined by my co-host, Eric Thornburg. Hello, and welcome back to The Cognitive Revolution. Today, I'm excited to welcome back Aravind Srinivas, founder and CEO of Perplexity AI AI. Aravind first appeared on the show back in March, and since then, he and the Perplexity AI team have continued to impress, shipping updates at such a relentless pace, and delivering results which so dramatically outshine Google and Bing, that Perplexity AI has even started to appear as a comparative standard for accuracy in academic papers. Speaking for myself, I can definitely say that Perplexity AI has become one of the AI tools that I use nearly every day, marking the first time in the last 20 years that a new app has meaningfully displaced Google in my everyday workflow.
And notably, when I asked Replit's VP of AI, Michele Katosta, what other companies he'd add to my AI Live Players list, he suggested just one, Perplexity AI. I think this conversation shows why that could in fact be a very good call. While Perplexity AI is still only about one one-thousandth the size of Google, they are now serving millions of queries per day, and their ambition, huge from the beginning, continues to expand. When I asked Aravind in March if he was worried about Google and Bing cutting them off from search index access, he said that he hoped they wouldn't do that.
This time, he said that they are working around the clock to build out their own web crawler, their own search index, and, yes, their own LLMs.
It seems that Aravind now expects the big tech giants to recognize how generative AI startups could disrupt their core businesses, and to begin to raise the drawbridges that currently span their proverbial modes. And so he aims to achieve technology self-sufficiency before that happens, such that he can sustain product supremacy if and when it eventually happens.
That strategy is not for the made of heart, or for the modestly resourced, and not something I'd recommend to most application startups. But judging purely by their track record of the last six months, I give Perplexity AI a very real chance of success. Briefly, a couple quick housekeeping notes before we get into the episode. First, ahead of this recording, I invited listeners to submit questions for Aravind, and I wanted to thank listeners Sid Harth, Ravi Kumar, as well as one who identified himself simply as John, for some very thoughtful questions. I touched on as many as I could, but wasn't able to ask all of them as I only had an hour with Aravind. But I really do appreciate the questions, and definitely plan to invite audience participation again in the future. And second, we continue to get feedback on audio quality, and we're definitely working on it. We now send any guests that need one a USB microphone ahead of our recording, and we aim to consistently deliver top-notch audio quality going forward. In the meantime, as always, if you're finding value in the show, please do share it with your friends or post a review on Apple or Spotify, or just leave a comment on YouTube. I recently got an amazing email from a history professor in Canada who got over a major hump in one of his projects when he followed my suggestion from our September 26th episode to fine-tune GPT 3.5 Turbo on GPT-4 reasoning. And I also wanted to call out our frequent YouTube commenter, AI in Check, who said that I look much better without my silly hat.