Fireside Chat with Perplexity AI’s Aravind Srinivas Offers Look into the Future of AI

Last week the Polsky Center for Entrepreneurship and Innovation welcomed Aravind Srinivas, CEO and cofounder of Perplexity AI, for a fireside chat on the future of artificial intelligence, the acceleration of agent-based systems, and what it really takes to build in today’s competitive AI landscape.

The conversation was opened by Samir Mayekar, managing director of the Polsky Center, and moderated by Sanjog Misra, Professor of Marketing and Faculty Director of the Center for Applied AI.

Srinivas, who previously served as a research scientist and holds a PhD from the University of California, Berkeley, cofounded Perplexity in 2022 with a goal to build the internet’s most accurate AI engine.

Since then, the company has grown rapidly. Today, the platform responds to more than 10 billion queries a year and is valued at $20 billion.

From Search to Answers

Aravind Srinivas, CEO and cofounder of Perplexity AI

Srinivas began by reframing a basic premise of the internet: search should return answers, not ads.

Traditional search, he explained, is fundamentally a marketing function — “If you search ‘pizza near me,’ the result is often whoever paid the most, not what’s actually best.” Instead, Perplexity was built to provide the most accurate, well-formatted summary, with the ability to follow up, refine, and complete tasks.

“The word ‘answer’ doesn’t just mean a link,” he said. “It can be an explanation, a summary, or even taking an action. That’s why we think the term ‘assistant’ is the better framing.”

The company’s long-term ambition, he added, is the general-purpose, reliable personal assistant — one that removes friction from daily life and unlocks deeper curiosity in humans.

When asked about what excites him most about the future of AI’s, Srinivas had a clear response.

“Everyone asks for three predictions. I only have one: accuracy,” he said. “Everything else breaks without it.”

From task completion to research, an inaccurate system becomes useless. The internet is already full of noise — “human slop,” as he put it – SEO-driven content, irrelevant pages, and endless open tabs. The way to fix that is an AI that can distill, verify, and act reliably.

Accuracy, however, is only the foundation. As Srinivas explained, the next leap forward comes from context — AI that understands what users are doing in real time.

AI Agents and the Second Brain

Srinivas pushed back on hype around the word “agents,” noting that Perplexity’s team thinks in terms of systems grounded in real usage.

One such system is Comet, Perplexity’s browser. According to research from Perplexity, when users have AI living directly inside their browsing experience instead of a separate window, they ask 6 to 18 times more questions. With full context from email, calendar, and daily workflow, AI gives the feeling of a second brain – a system that can draft responses, organizes tasks, prioritizes what matters, and frees time for deeper thinking.

“It’s not just productivity. Curiosity is uniquely human,” he said. “Workflows get in the way of exploration. If an assistant handles the workflows, people can focus on what interests them.”

Rather than make bold predictions on what’s next, Srinivas cited Linux creator Linus Torvalds: great systems are built by fixing thousands of tiny problems.

“The dream assistant — one that is booking travel, scheduling doctor’s appointments, making purchases — that sounds magical,” he said. “But to make that real, you need reliability. That comes from fixing the thousand bugs no one wants to fix.”

That reliability is also what will define the next phase of business models in AI.

Unlike traditional ads-driven search, Perplexity is looking at subscriptions and usage-based value.

If an AI saves a user money, finds discounts, or completes transactions, that value can be shared among the user, the agent, and the business. He predicts that consumer AI can achieve strong margins — maybe not 90% like some search engines, but strong enough to build healthy companies.

“But first the assistant has to be reliable, accurate, and trustworthy,” he said. “We can always monetize value. The question is: are we providing value no one can live without? Not yet. No AI is there.”

Advice for Students interested in Entrepreneurship

For students and founders exploring AI, he ended with one theme: grit.

“I’d have to be dead or incapacitated before I give up,” he said. “Every founder hits roadblocks. The difference is who keeps going.”

His closing message: “Stay hungry, stay curious and listen to your gut. Don’t do things a certain way just because that’s how others do it – find a different way or a new angle to an overlooked path.”

“And most importantly – never give up,” he continued. “That’s the number one quality of every successful entrepreneur.”

Want to join future Polsky Center events? Follow us on LinkedIn and Instagram, and subscribe to our newsletters to stay connected.