Overview

Perplexity AI is an American artificial intelligence company founded in August 2022 and headquartered in San Francisco, California. The company develops a conversational search engine that generates direct, cited answers to natural-language questions, positioning itself as an alternative to traditional list-of-links search. It was co-founded by former staff of OpenAI, Meta, and Databricks, and its flagship product launched in December 2022, one week after ChatGPT.

Company Profile

Founding Background

In the fall of 2022, Aravind Srinivas left OpenAI, where he had worked after completing a PhD at UC Berkeley, to start the company with Denis Yarats, Johnny Ho, and Andy Konwinski. The name "Perplexity" borrows a machine-learning term measuring how uncertain a language model is about its next prediction, reflecting the company's aim of reducing that uncertainty for users seeking information. The entity was formally incorporated in August 2022.

Headquarters and Team

The company is based in San Francisco. The four co-founders bring complementary backgrounds: Aravind Srinivas (CEO) previously researched AI at OpenAI, Google Brain, and DeepMind; Denis Yarats (CTO) worked at Meta AI; Johnny Ho (CSO) was an engineer at Quora and a quantitative trader at Tower Research Capital; Andy Konwinski co-founded Databricks.

Products and Services

Perplexity AI Search Engine

Product Form

The core product is a conversational search engine. Users ask questions in natural language, and the system retrieves web sources and synthesizes a cited summary rather than returning a ranked list of hyperlinks. It is available on the web, as mobile apps, and as browser extensions.

Citations and Sourcing

Each key claim in an answer carries a numbered citation linking back to the original web page, so users can verify the underlying source. The company also operates a web crawler called PerplexityBot, which it states is used to build the search index for retrieval and citation, not to train foundation models.

Perplexity Pro Subscription

Perplexity Pro is a paid consumer subscription offering more frequent access to frontier models, longer context, file uploads, and related premium features. It is one of the company's principal revenue streams.

Comet Browser

In 2025 the company released Comet, an AI browser that embeds Perplexity search and question-answering directly into a sidebar, pitching an agent-driven model in which the assistant browses and acts on web pages on the user's behalf.

Business Model

Subscription Revenue

Revenue comes primarily from paid subscribers to Perplexity Pro and related plans. Annualized revenue was roughly 50 million USD in early 2024, and by 2026 public reporting placed it around 500 million USD.

Advertising and Partnerships

Beyond subscriptions, the company has experimented with brand advertising such as sponsored follow-up questions, and pursued content-licensing partnerships with publishers as a way to share value between generative search and content creators.

Technical Approach

Model Selection Strategy

Perplexity does not train its own foundation models. Instead it calls third-party models such as GPT, Claude, and Llama through APIs, and builds its retrieval-augmented generation (RAG) pipeline and product experience on top. This lets it adopt the newest models quickly but creates upstream cost and dependency risk.

Real-Time Retrieval

Answers are grounded in live web retrieval. The system fetches and ranks multiple sources before a language model synthesizes the final response, mitigating the knowledge-cutoff limitation of purely parametric models.

Growth and Controversies

Funding History

The company raised capital at a rapid pace: in January 2024 it closed roughly 74 million USD at a valuation near 520 million; by March 2024 it was valued at over 1 billion; in June 2024 a SoftBank Vision Fund 2 investment pushed the valuation to 3 billion; in December 2024 a 500-million-USD round led by Institutional Venture Partners valued it at 9 billion; and by 2025 its valuation reached approximately 20 billion USD.

In October 2024, News Corp-owned Dow Jones and the New York Post sued Perplexity in the U.S. District Court for the Southern District of New York, alleging that it copied large volumes of copyrighted news content into its RAG index to generate answers, a practice they described as "massive freeriding" that diverted readers and advertising revenue from publisher sites. In August 2025 the court denied Perplexity's motion to dismiss for lack of jurisdiction or to transfer the case to California, and litigation continued in New York. Other outlets, including The New York Times and Forbes, have also publicly accused Perplexity of unauthorized use or of hallucinated misattribution. Perplexity has argued that its system relies on publicly available factual information, that it has anti-infringement guardrails in place, and that the publishers had tried to induce its model into verbatim reproduction before suing.

Trademark Claims

The publishers additionally alleged that Perplexity attributed hallucinated, fabricated passages of news to real newspaper names, which they argued constitutes trademark dilution and false designation of origin under the Lanham Act.

See Also

  • Perplexity
  • Retrieval-augmented generation
  • Large language model
  • Conversational search