Jul 11, 2025 · 43m · y-combinator
Aravind Srinivas: Perplexity's Race to Build Agentic Search · Y Combinator
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Y Combinator AI Startup School fireside chat, Perplexity CEO Aravind Srinivas joins General Partner David Lieb to discuss Perplexity's growth, its strategic push toward an agentic AI browser, and strategies for competing against tech incumbents. Srinivas shares insights on finding product-market fit, maintaining high execution velocity, and cultivating founder resilience in the rapidly evolving AI landscape.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The partners as informed peer 3.5 Guest teaching 4.8 Guest disagreement 2.0 The partners pushing back 1.9
speaking balance: gold is the partners, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 6:07 Challenging canonical Y Combinator startup advice
Aravind openly rejects standard YC dogma by declaring that Perplexity began without a fixed product idea, urging founders not to bind themselves rigidly to a single concept in fast-moving AI.
Hardest push from the partners ▶ 40:43 Host presses on accuracy benchmarking for subjective queriesDavid Lieb directly challenges Aravind on how Perplexity validates answer quality when there is no single objective truth, pushing past generalities to extract an admission that systematic evals are not yet in place.
Biggest teaching moment ▶ 12:14 Masterclass on Google's advertising dilemmaAravind thoroughly educates the room on how Google's multi-billion dollar bidding auctions from travel and commerce aggregators structurally prevent them from surfacing direct, concise AI answers.
The partners hold their own ▶ 2:53 Host frames the looming multi-front AI search collisionDavid Lieb demonstrates acute industry insight by framing how Sam Altman and Sundar Pichai are aiming at the exact same product surface, forcing Aravind to articulate his core defensive thesis.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The partners as informed peer | Guest teaching | Guest disagreement | The partners pushing back | Why |
|---|---|---|---|---|---|---|
| Y Combinator Batch Application Announcement | 2 | 4 | 1 | 1 | David Lieb opens the session by warmly welcoming Aravind Srinivas and asking for an overview of Perplexity's current state and traction. Aravind details their explosive infra scaling challenges and lays out their strategic shift toward building an agentic cognitive browser. | |
| Navigating Competition with AI Giants and Speed as a Moat | 5 | 4 | 2 | 4 | Lieb poses a pointed strategic challenge, observing that OpenAI and Google are heading toward the exact same product destination. Aravind responds pragmatically, arguing that high-value opportunities will always attract well-funded copycats and that startup speed is the sole viable moat. | |
| Hands-On CEO Leadership and Bug Fixing | 3 | 5 | 3 | 1 | Lieb highlights Aravind's hands-on bug-fixing mindset witnessed backstage. In sharing Perplexity's founding story, Aravind playfully contradicts standard YC doctrine by explaining that they started without a specific idea or project. | |
| Academic Roots, Co-Founders, and Intellectual Humility | 3 | 4 | 1 | 1 | Aravind shares the progression from academic research and early SQL/Twitter search tools to launching Perplexity right after ChatGPT. He emphasizes the importance of intellectual humility in choosing problems tailored to founder strengths. | |
| Recognizing Google's Innovator's Dilemma | 4 | 6 | 2 | 1 | Aravind provides a sharp breakdown of Google's Innovator's Dilemma, explaining how direct answers dismantle high-margin CPC bidding models from travel and shopping aggregators. Lieb concurs with the strategic analysis. | |
| Corporate Inertia versus Startup Agility | 3 | 5 | 3 | 1 | Aravind candidly critiques Google's repeated announcements of AI search features under changing branding without broad rollouts. He notes how organizational inertia and stock sensitivity open structural windows for startup disruption. | |
| Competing with OpenAI and the Browser Abstraction Layer | 3 | 5 | 2 | 2 | Addressing competition with OpenAI, Aravind describes the Comet browser as an abstraction layer positioned above individual chatbots. He argues that deep OS and browsing integrations will be difficult for pure model providers to replicate swiftly. | |
| Reimagining the User Experience and Agentic Workflows | 4 | 5 | 2 | 3 | Lieb presses for concrete use cases that would persuade Chrome users to switch. Aravind illustrates multi-step agentic workflows and later explains how Perplexity mandates AI coding tools internally while warning against 'vibe coding' core infrastructure. | |
| Defensibility, Brand Identity, and Accuracy Focus | 4 | 6 | 2 | 2 | Lieb asks how Perplexity maintains defensibility as AI coding tools lower barriers to entry. Aravind argues that brand identity and accuracy are durable assets, and contrasts Perplexity's native browser agent approach with brittle third-party MCP protocols. | |
| Business Models: Subscriptions, Usage Pricing, and Transactions | 4 | 5 | 2 | 2 | Lieb inquires about revenue potential relative to Google's search monopoly. Aravind candidly concedes Perplexity may never match Google's historic ad margins, outlining a hybrid model of subscriptions, usage-based compute pricing, and transaction fees. | |
| Advice for Founders: Relentless Execution and Embracing Existential Fear | 3 | 4 | 2 | 1 | Asked for founder advice, Aravind urges entrepreneurs to expect well-funded model labs to clone profitable ideas and to embrace existential fear rather than over-strategizing defensively. | |
| Audience Q&A: Facing Google's Dilemma and User Resistance | 3 | 4 | 2 | 2 | Audience members ask about scaling partnerships, overcoming startup failure, and competing when LLMs integrate search. Aravind emphasizes persistent iteration, niche mastery, and resilience inspired by Elon Musk. | |
| Audience Q&A: The Impact of AI Search on Web Traffic and Publishers | 5 | 5 | 2 | 4 | Audience members raise concerns regarding web scraping ethics, power-law traffic concentration, and political bias. Lieb steps in to ask how Perplexity measures subjective accuracy, prompting Aravind to admit that automated eval frameworks for subjective queries remain an unsolved challenge. |
Statements from this episode (38)
Srinivas: Perplexity faces daily infrastructure issues due to rapid usage growth
“Well, whether you believe it or not, like, I have infra issues every day, so there are a lot of people using it, and this usage is actually growing to the extent that we don't actually know how to deal with it. We have to rebuild the infra to scale the next NX…”
Srinivas: Perplexity's primary strategic bet for the future is an AI browser
“The browser. That's the big bet we are making as far as the future of the company goes.”
Srinivas: AI browsers should be assistants combining navigation, search, and agents
“We think about it as an assistant rather than a complete autonomous agent, but one omnibox where you can navigate, you can ask informational queries, and you can give agentic tasks, and your AI with you on your new tab page, on your sidecar as an assistant on …”
Srinivas: An AI browser is much harder to copy than chat tools
“The browser is much harder to copy than, like yet another chat tool.”
Srinivas: OpenAI and Anthropic Will Both Build Their Own Browsers
“That said, I'm fully working with the assumption that OpenAI will also build its own browser. Anthropic will also try to build its own browser.”
Srinivas: Speed is a startup's only defensible moat against AI incumbents
“And the only mode you have is speed. You have to innovate. You have to move faster than everybody else.”
Srinivas: Sundar Pichai is personally doing bug support on X
“Like I've noticed even Sundar is doing bug support on X right now.”
Srinivas: Perplexity started without a clear idea, defying YC advice
“We started the company without actually having clear idea of what to build, which is the opposite of what YC advice is, which is start from a project and turn it into a company.”
Srinivas: There is still no good way to search X
“There still is no good way to search over Twitter.”
Srinivas: Perplexity bet LLMs would handle reasoning over unstructured data
“So we bet on the fact that language models can do all the reasoning and parsing and like structuring later, but the more important thing is to start with something more unstructured, and that ended up becoming perplexity.”
Srinivas: Perplexity launched seven days after ChatGPT, which lacked web search
“We launched it seven days after the ChatGPT launch, especially at a time when ChatGPT did not have web search.”
Srinivas: Google's ad model disincentivizes providing direct search answers
“If you get direct answers to these questions with booking links right there, how are you going to mint money from booking and Expedia and Kayak and like, you know, all like same, same, same thing for shopping. How are you going to take money from Amazon and li…”
Srinivas: Google possessed only fourth or fifth best AI models in 2023-2024
“And 2024, 20 23 especially, and large part of 2024 too, Google had like, maybe a fourth or fifth best models at any moment. So as a startup outside Google, you had access to AI that was better than what Google internally had.”
Srinivas: Google rebrands and announces the same AI search features every year
“The same feature is being launched year after year after year with a different name, with a different VP, with a different group of people, but it's the same thing. Except maybe it's getting better, but it's never getting launched to everybody.”
Srinivas: Phone Makers Will Offer Multiple AI Search Alternatives
“The phone makers will start offering all of them as alternatives. It's not going to be like a locked-in default search option.”
Srinivas: ChatGPT's dominance prompted Perplexity to pivot to building a browser
“ChatGPT is the most successful consumer AI product out there, and so competing against it is very difficult, which is why I really want to, like, push the company more on the browser side, and I think Comet, the browser, will be an abstraction layer about chat…”
Srinivas: Vibe coding cannot replace traditional infrastructure and distributed systems engineering
“Like we frequently run into infra issues, and you don't want a vibe coder right there fixing it on live things on production. Like, I do want, like, people well trained in regular software engineering, infrastructure, distributed systems. Like, you don't want …”
Srinivas: Perplexity mandates that engineers use AI tools like Cursor or Copilot
“We made it mandatory to use at least one AI coding tool, and internally at perplexity it happens to be cursor and like a mix, mix of cursor and GitHub Copilot. But yeah, we definitely made it compulsory.”
Srinivas: Claude Code appears to be far smarter than Cursor
“Claude code seems to be far smarter than, like, what Cursor is able to do.”
Srinivas: OpenAI's built-in search features in ChatGPT did not kill Perplexity
“OpenAI has perplexity within ChatGPT that did not kill any of these companies.”
Srinivas: Millions of paying users provide startups brand durability against incumbents
“So there is a certain brand value that once you acquire at the scale of, like, several millions of users, paying users, you don't actually die that fast.
[1380] Aravind Srinivas: You earn the right to survive and keep building.”
Srinivas: Perplexity maintains the fastest time-to-first-token despite performing live web searches
“We focus on speed, time to first token on app or web, like we're still the fastest despite doing search.”
Srinivas: No AI product has in-app network effects yet
“No AI product has within app network effect. Like it's not like WhatsApp where if you build a WhatsApp rival, Meta has a definitely like a questionable brand, right? Like people don't necessarily trust Meta's products. They think like these are ad products. De…”
Srinivas: Perplexity partners with Shopify, Yelp, and TripAdvisor for vertical search
“We already work with self book. They power all the hotel bookings natively done on perplexity. We work at TripAdvisor to surface all the reviews of hotels and Different, you know, places. We have, like, collaborations for the maps. We work with Yelp. We also, …”
Srinivas: Browser-native agents offer superior reliability over MCP-only approaches
“That's the key advantage of the browser that you do not have if you commit entirely to just the MCP vision. If you commit entirely to the MCP vision, you require these third-party MCP servers to work reliably. Ah, the data that they send you, ah, on, with the …”
Srinivas: Perplexity can reach billions annually in subscription revenue alone
“We think, like, we can grow at least, you know, a few billions a year in just subs, which is a great business.”
Srinivas: Task-based pricing for AI agents will become a major model
“Usage-based pricing. Where people are paying an agent for completing a task, or people have recurring tasks, and they pay based on, you know, every single use of the task, and they normalize this system based on, like, how much it would take to hire a person t…”
Srinivas: Google avoided transactions because CPA margins trail CPCs
“CPAs have historically been way lower margins than CPCs, which is why Google never became a transaction platform.”
Srinivas: Foundation model labs will copy any high-revenue AI startup idea
“You should assume that if you have a big hit, if your company is something that can make revenue on the scale of hundreds of millions of dollars, or potentially billions of dollars, you should always assume that a model company will copy it. Mainly because the…”
Srinivas: Forcing AI on massive mainstream user bases is difficult
“There are a lot of people in the world who hate AI, by the way, so I think just throwing AI down people's throats on such a, you know, massive distribution area is not easy.”
Srinivas: Users will hate AI answers if ads are embedded into them
“If ads are part of every AI answer, you're going to hate it too.”
Srinivas: Browsers are the natural graduation step from search engines
“And browser and search are not two distinctive products. They're actually, like, the browser is a natural graduation step from search. Just like how Google graduated from Google search to Chrome. And Chrome is the main reason they got billions of daily queries…”
Srinivas: Autonomous mobile AI agents can only be built inside web browsers
“And agents can only be built with a browser. I, I'm very, like, convinced about that vision, that if you want to have a mobile agent, That you can actually build and implement without being restricted by whatever OS rules that Apple or Google sets in terms of …”
Srinivas: Mobile apps will resist adopting MCP servers for AI
“Expecting every mobile app to have MCP servers and then, like, connecting all their data to your thing, It's not going to be that straightforward. Like, nobody wants to be disintermediated by an AI that quickly.”
Srinivas: Web traffic power law will become even more skewed
“I think that there are going to be, ah, you know, the web is already pretty long tail, ah, and there's a massive power law, so I feel like the power law is going to get even more skewed. That is very obvious.”
Srinivas: SEO-gaming websites will definitely have a harder time
“But those who are trying to game the SEO system and trying to get traffic, I think they're definitely going to have a harder time.”
Srinivas: AI search should offer all perspectives on subjective topics
“Now there are things that don't have one clear accurate answer. I think they're the best thing we can do is offer all the perspectives. And not really take a clear opinion on like what is right and wrong when there's no clear answer to that question.”