YC
How To Build The Future: Aravind Srinivas
@ Y Combinator
02/22/25
Takeaways
AS
Aravind Srinivas
02/22/25
@ Y Combinator
We released the ability to ask follow-up questions, which doubled engagement time on the site and increased the number of questions asked daily.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
Perplexity is designed to be a more intelligent search engine than Google, particularly useful in specific scenarios.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I learned from my internship at OpenAI that the only research that truly matters is in unsupervised learning and reinforcement learning.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I realized that if people stop clicking on links, the ad economy dies, which was a key insight we had after launching our product.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
The motivation behind starting Perplexity wasn't to kill Google, but to explore an idea of significant scale and ambition.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
We initially focused on searching over specific verticals or datasets, but pivoted to a more general search approach when we realized the limitations of that strategy.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
The first real product we launched was a chat UI that allowed users to converse and retrieve data from a database, which was well-received.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
We transitioned from a Twitter search demo to building Perplexity by recognizing the potential for a more general solution that could index various databases.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I realized there was potential in Perplexity when I saw the exponential increase in user engagement and questions being asked.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I was initially more concerned about competition from Microsoft and Bing chat rather than Google, as I believed our product addressed a unique need.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I deliberately designed our product with attention to detail, ensuring that I am the primary user, which helps in creating a better experience.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
The user is never wrong, and even if a feature doesn't work, it's important to clarify the user's intent rather than blame them.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
Our primary metric is the number of queries per day, and we analyze weekly and monthly growth rates to understand our performance better.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
To avoid becoming a big, slow company, we focus on maintaining a culture where everyone feels comfortable raising bugs and discussing issues directly.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
Perplexity aims to be an all-in-one experience where users can find answers and fulfill actions, like purchasing products directly from search results.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
The challenge lies in balancing the need for accurate information with the user experience, ensuring that users can quickly find what they need without unnecessary complexity.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I think the obsession with user experience and good product taste will give us an edge over competitors like Google, who may prioritize ads over user needs.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
I believe that the next generation of search will require a new monetization model that effectively addresses user needs while providing a seamless experience.
AS
Aravind Srinivas
02/22/25
@ Y Combinator
We are capable of utilizing the latest open-source models and fine-tuning them to create a better information experience for users.