What is Perplexity AI? Definition, history and how it works

How does Perplexity AI work?

Perplexity works by running a live web search for each question, reading the best pages and writing one answer with links to them. Instead of a list of ten blue links, the user gets a synthesised summary and can click each citation to check the original page. The process has four steps:

1. Interpret the question and break it into search queries (a pattern known as query fan-out).
2. Retrieve fresh pages from the web through Perplexity’s own index and crawlers.
3. Generate an answer with a large language model chosen by Perplexity or by the user.
4. Attach numbered citations so the reader can verify every claim.

Paying users can pick the underlying model, including OpenAI GPT and Anthropic Claude models, and Perplexity also runs its own Sonar models. The Help Center lists modes such as Pro Search and Deep Research for more complex questions.

History of Perplexity AI

Perplexity was founded in August 2022 by Aravind Srinivas, Denis Yarats, Johnny Ho and Andy Konwinski. The table shows the key milestones.

Perplexity has also faced copyright disputes, including a lawsuit from Dow Jones and the New York Post and suits from Japanese publishers Yomiuri Shimbun, Asahi Shimbun and Nikkei in 2025.

Perplexity AI vs Google Search vs ChatGPT

Perplexity sits between a classic search engine and a chatbot: it answers like ChatGPT but always grounds the answer in live web results, like Google.

Perplexity AI in SEO and GEO

Perplexity matters for SEO because every answer links out to the pages it used, so being cited sends referral traffic. This is the core of Generative Engine Optimization (GEO): you want your page to be one of the few sources an AI engine quotes.

Crawler access

Perplexity documents two user agents. PerplexityBot indexes pages to show and link them in results and, according to Perplexity, is not used to train foundation models. Perplexity-User fetches a page when a user asks a question and, as its docs state, generally ignores robots.txt because the fetch is user-initiated. Block PerplexityBot in robots.txt only if you accept losing visibility in its answers.

Content that gets cited

• Answer the question in the first sentence under each heading.
• Name entities, dates and numbers with their source.
• Keep facts on crawlable HTML pages, not only in PDFs or images.
• Build brand mentions on sites Perplexity already trusts, a topic I cover in the entity SEO guide.

If you want this audit done for your brand across several markets, see my AI SEO services.