Y Combinator Startup Podcast How To Build The Future: Aravind Srinivas
137 snips
Feb 21, 2025 
Aravind Srinivas, co-founder and CEO of Perplexity, shares his journey from an intern at OpenAI to leading a new AI-powered search engine. He explores how to compete with giants like Google, emphasizing the need for innovative user experiences. Discussions include the integration of AI in search, the importance of passionate hiring, and the challenges of scaling a tech startup. Aravind also highlights the role of user retention and feedback in shaping future products, paving the way for a new era in search technology.
AI Snips
Chapters
Books
Transcript
Episode notes
Early Pitch and Pivot
- Aravind Srinivas pitched disrupting Google through a glasses interface, focusing on visual search.
- Investors advised against it, leading to a focus on searching specific data sets like Twitter.
General vs. Domain-Specific Search
- Building a general solution with unstructured data and relying on LLMs is more scalable than domain-specific indexing.
- This approach offers an advantage against legacy systems like Google's structured approach.
The "Dumb Approach"
- Perplexity adopted a "dumb approach" by feeding top search results and snippets directly to the LLM.
- This simpler method, combined with improved LLMs like GPT-3.5 Turbo, proved surprisingly effective and faster.
