Perplexity

Perplexity sits between a search engine and an AI assistant. Where a search engine returns a list of links and an assistant often answers from memory, Perplexity is built around searching first and showing its sources by default. It is one of the clearest examples of an answer engine.

The company behind it, Perplexity AI, was founded in 2022 and is based in San Francisco. It uses a mix of its own models and models from other AI companies, so the product is defined more by its search-and-cite approach than by a single model. Its answers depend on what its search finds on the day.

How it works

When you ask a question, Perplexity first turns it into one or more search queries and gathers relevant pages from the web. It reads the most useful passages and passes them to a large language model, which writes an answer based on that material. Each claim is tagged with a number linking to the page it came from, a form of grounding.

Follow-up questions keep the earlier context, so you can narrow down a topic the way you would with a person. Some modes run many searches in turn and write a longer, report-style answer.

Perplexity documents its crawler, PerplexityBot, which follows robots.txt, and a separate agent that fetches a page when a user asks about it, which robots.txt generally does not govern. A page Perplexity cannot reach is unlikely to be cited. Because the answer is written from whatever the search returns, the same question asked twice can cite different sources, especially for local or fast-changing topics.

Example

Someone asks Perplexity which landscapers near them build dry-stone walls. The answer names three firms, each with a citation. Two citations lead to the firms' own service pages, which describe dry-stone walling in plain words with photos and a service area. The third leads to a directory review. A firm that offers the service but only mentions it in a photo caption is not named.

The owner of that firm checks by asking the same question a few ways, noting which pages are cited. He adds a dedicated page about dry-stone walling, with the towns he covers, and tests again after it has been indexed. He also checks that the new page states his business name, the towns he serves and how to get a quote in the opening lines, since answer engines often quote from the start of a page. A month later the new page is cited for two of his test questions, though not all, which is a realistic result: citations move with every update to the index.

Why it matters for your business

Perplexity's citations show plainly what kind of page gets used in AI answers: specific, clearly written pages that answer one question well. That makes it a useful free test for your own content, even if few of your customers use it. Being cited can send visitors who are close to deciding, but traffic from it is usually small next to Google, so treat it as one channel, not a strategy. Checking it also costs nothing: ask your customers' typical questions and click each citation to see which page won and why.

Common mistakes

• Hiding key services inside images or PDFs where an answer engine cannot easily read them.
• Blocking AI crawlers by default and then wondering why the business is never cited.
• Testing with a single question and drawing conclusions, when answers vary with wording.
• Assuming a citation in Perplexity means the same page ranks well in Google, or the reverse.

Questions people ask

Who makes Perplexity?

Perplexity is made by Perplexity AI, a company founded in 2022 and based in San Francisco. It offers the answer engine on the web and in apps, with free and paid plans, and uses a combination of its own models and models from other AI companies.

How is Perplexity different from ChatGPT?

Perplexity is built around web search and shows citations for its answers by default, so it behaves like a search engine that writes summaries. ChatGPT is a general assistant that can also search but is equally used for writing, analysis and other tasks that need no search.

How do I get my site cited by Perplexity?

Make sure its crawler can reach your pages, then publish clear, specific pages that answer the questions your customers ask, with your location and services in plain text. Consistent details across directories and reviews help. Test by asking real customer questions and noting which pages it cites.