“We want to be the world's most knowledgeable and trustworthy source of information.”
The Story
Web search has worked the same way since 1998: enter a query, receive a ranked list of links, visit the links to find answers. The ranking algorithm changed radically over those decades. The interaction model did not. You still had to do the reading.
Perplexity was founded in 2022 on the observation that the link list was an artifact of what was technically possible in 1998, not of what was actually useful. The question a user types into a search box is almost never "show me ten pages that might contain an answer." It is "tell me the answer." Large language models made that possible. Retrieval-augmented generation made it grounded and citable.
The product Perplexity built — which it calls an "answer engine" — takes a query, searches the live web in real time, retrieves relevant sources, and synthesizes a direct answer with inline citations. The citations are not decorative; they are linked and verifiable. The answer is not the model's training data frozen at a cutoff date; it is a synthesis of documents retrieved seconds ago.
By 2025, Perplexity was handling 100 million weekly queries. The company had become the clearest demonstration that the link-list paradigm was not inevitable.
Why They're in the Hall
Perplexity belongs in the Hall for two reasons that sit in productive tension.
The first is the achievement: they built the first AI-native search product that people actually use at scale, demonstrating that retrieval-augmented generation works in production and that users prefer direct answers to ranked links when the answers are reliably sourced.
The second is the failure mode they embody: a system that synthesizes and compresses web content into direct answers reduces the traffic that flows to the original sources. The journalism, the documentation, the technical writing that Perplexity cites is funded by the traffic its answer engine redirects. This is a real attribution and sustainability problem — not unique to Perplexity, but most visible there because they are most explicit about citation. The irony is precise: the more accurately they cite sources, the more clearly the traffic is not flowing to them.
The Pattern
Perplexity is a Transitive Trust case study at the information layer: the user trusts the answer engine, the answer engine trusts the retrieved sources, the retrieved sources were created by publishers who trusted that search traffic would fund their continued creation. Perplexity changes the middle term in that chain without changing the endpoints. The trust propagates; the traffic does not.
They solved the wrong half of the problem with perfect technical elegance. The right half — who pays for the content being synthesized — is still open.