How Did Aravind Srinivas Build Perplexity AI into a Fast-Growing Conversational Answer Engine?

TL;DR

Perplexity AI grew rapidly by replacing traditional lists of search links with direct, cited answers to natural-language questions. Launched on December 7th, 2022, it reached about 10 million monthly active users and began serving more than 3 to 4 million queries a day. Aravind Srinivas explains how academic research, generative AI, reinforcement learning, and interactive clarification shaped the product. Read on for the principles behind its search experience and growth.

Transcript

I'm Arvind Srinivas. I'm the co-founder and CEO of perplexity AI. Perplexity is a conversational and search engine that aims to deliver answers to you, to whatever questions you may ask. We are trying to revolutionize how people consume information online. Instead of getting ten blue links, they can just ask questions in natural language and just g... Read More

Key Insights

• Perplexity AI aims to revolutionize online information consumption by providing direct answers to user queries instead of traditional search results.
• The startup achieved 10 million monthly active users within a year, highlighting its rapid growth and widespread adoption.
• Aravind Srinivas, the co-founder, transitioned from academia to entrepreneurship, driven by a desire to execute his own vision.
• Perplexity AI combines generative AI and reinforcement learning to create a conversational search engine, offering an innovative user experience.
• The startup's pricing strategy aligns with ChatGPT Plus to ensure users value its unique offering of combining LLM and search.
• Perplexity AI emphasizes the importance of focusing on a few high-quality features to maintain user trust and satisfaction.
• The company's culture encourages urgency and strategic decision-making, ensuring efficient execution and innovation.
• Aravind stresses the importance of passion in entrepreneurship, focusing on improving the product rather than just financial metrics.

Questions & Answers

Q: What is Perplexity AI, and how does it differ from a traditional search engine?

Perplexity AI is a conversational answer engine that responds directly to questions asked in natural language. Instead of presenting ten blue links for users to open and interpret, it generates an answer and provides a corresponding reference, or citation, for each sentence.

Q: How quickly did Perplexity AI grow after launching?

Perplexity AI launched on December 7th, 2022 and received around 2000 to 3000 queries on its first day. It later reached about 10 million monthly active users and began serving more than 3 to 4 million queries a day, representing thousand X growth over one year.

Q: Why does Perplexity AI include citations in its answers?

Aravind Srinivas and his co-founder Dennis brought the practice from their academic backgrounds as PhDs. Just as claims in academic papers must be supported by references, Perplexity connects its generated statements to sources so its responses resemble researched writing.

Q: What does Perplexity Copilot do?

Copilot asks clarifying questions when a user begins with an incomplete or uncertain prompt. It interactively expands the request so the AI can better understand what the user wants without expecting the user to be an expert prompt engineer.

Q: How did Aravind Srinivas become interested in artificial intelligence?

While studying at an IIT in India, Srinivas entered a machine-learning contest after a friend introduced him to the problem of predicting outputs from a dataset. He won the contest, found the work came naturally, and decided to study the field more deeply before completing a PhD at Berkeley in AI and deep learning.

Q: How did Aravind Srinivas's OpenAI internship influence his development?

During his research internship at OpenAI in 2018, Srinivas realized that he needed to improve his programming, first-principles thinking, and clarity of thought. The publication of GPT 1 also convinced him that learning from internet data would become increasingly important.

Q: What research helped shape the technology behind Perplexity AI?

Srinivas and his advisor spent two years learning and coding around emerging approaches based on internet data. This work led him to study how generative AI and reinforcement learning could be combined to produce systems that both predict language and communicate effectively with humans.

Q: Why did Aravind Srinivas choose entrepreneurship over an academic career?

Srinivas saw becoming a professor or an entrepreneur as the paths that would let him execute his own vision. Reading How Google Works and reflecting on the possibility of an AI-powered ultimate search engine helped him connect that ambition with entrepreneurship.

Summary & Key Takeaways

• Perplexity AI is revolutionizing the way people interact with online information by providing direct answers to user queries, rather than traditional search results. Founded by Aravind Srinivas, the startup has achieved rapid growth, reaching 10 million monthly active users within a year of its launch.

• Aravind Srinivas, formerly an academic, founded Perplexity AI to execute his vision of a conversational search engine. The startup combines generative AI with reinforcement learning to offer an innovative user experience, challenging traditional search engines and gaining significant traction.

• Perplexity AI's strategic pricing and focus on high-quality features have contributed to its success. The company emphasizes urgency and strategic decision-making, ensuring efficient execution and continuous improvement. Aravind believes passion and mission-driven goals are crucial for entrepreneurial success.