‘You can only be world-class at one thing’: Why Perplexity CEO thinks speed, focus and not funding may beat Google

At Y Combinator’s AI Startup School, Aravind Srinivas, CEO of Perplexity, explained why Big Tech can’t win at everything—and why Perplexity’s AI browser bet is harder to copy than ChatGPT.

“Perplexity isn’t trying to build the most general AI, it’s trying to be the best at one thing: factual, fast, and reliable answers. Everything else, including a browser, is in service of that,” he told the gathering.

Addressing aspiring entrepreneurs, Srinivas said the AI race is heating up but smaller startups can outpace Big Tech not by outspending them, but by out-focusing them. He noted that while companies like Google or OpenAI have deep pockets, “they can’t afford to fail publicly.”

This, he said, was a case of focus versus scale, speed versus structure, and mission-driven product thinking versus distribution-first monetisation.

Innovator’s advantage vs big tech’s burden

Srinivas argued that even if tech giants tried to replicate Perplexity’s approach, they wouldn’t be able to do so without taking on enormous risk. “Google and OpenAI can try to copy Perplexity, but they can’t afford to fail publicly.”

He added, “Building a browser with real agentic abilities is much harder than spinning up another chatbot.”

Perplexity’s lean infrastructure and UX-first mindset, he said, allow for faster execution. “Perplexity’s lean infra and obsessive UX culture allow for faster iteration—like a high-speed marathon.”

Comet browser as an operating system for knowledge

Perplexity’s long-term vision goes beyond AI-powered search, Srinivas said. The company’s Comet browser aims to reinvent the browsing experience entirely. “Chrome was exciting when each tab was its own process. With Comet, each query or task becomes its own process. That’s the leap.”

Rather than tabs, Comet organises workflows into “asynchronous cognitive tasks”—meant to function like embedded research assistants across a user’s digital activity.

Why Perplexity won’t die even if ChatGPT copies 

Even if rivals mimic Perplexity’s features, Srinivas believes core elements like brand, trust, and execution speed can’t be cloned. “Brand and obsessive focus (on things like time-to-first-token and reliable citations) matter more than people think," he said adding that  speed of iteration and trust built over time are harder moats than scale or model quality alone.

One bug at a time

Srinivas also pointed to Perplexity’s hands-on culture—saying he still personally fixes bugs despite being CEO. “Is that the best use of a CEO’s time? Maybe not. But it shows where our priorities are.”

He concluded: “There’s only a limited number of things you can be world-class at. For us, it’s answering questions with extreme accuracy, orchestrating agentic tasks, and now reimagining the browser.”