Perplexity is designed to be a more intelligent search engine than Google, particularly useful in specific scenarios.

I realized there was potential in Perplexity when I saw the exponential increase in user engagement and questions being asked.

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.

For clients like Perplexity, the challenge is appealing to a broad audience, making it aspirational and relatable at the same time, while also transforming the AI experience from a simple question-and-answer interaction into a comprehensive knowledge portal, similar to how Google used to serve as the front page of the internet.

The key differentiator between Perplexity and chatbots is that while chatbots provide direct answers, Perplexity opens doors to further knowledge and discovery.

Perplexity aims to be an all-in-one experience where users can find answers and fulfill actions, like purchasing products directly from search results.

The motivation behind starting Perplexity wasn't to kill Google, but to explore an idea of significant scale and ambition.

Transformers, particularly BERT and MUM, have significantly improved search quality, leading to new use cases and query types.

AI excels at semantic search, allowing it to find relevant information from arbitrary text, which was challenging before.