What it means
Perplexity describes itself as an answer engine. For a typical question it searches the web, selects sources and has a language model write the answer, with numbered citations beside the claims they support. It can use several language models.
Its search crawler is PerplexityBot, which Perplexity says surfaces and links websites in its results and is not used to crawl content for AI foundation models. A separate agent, Perplexity-User, visits pages when a user's question needs them, and Perplexity says it generally ignores robots.txt for those requests.
Why it matters
Because Perplexity shows its sources prominently, the evidence behind a recommendation is easy to inspect. That makes it a practical place to see which pages, reviews and comparisons shape answers in your category, and whether your own site is among them.
Example
Asked which customer data platforms suit a mid-market B2B company, Perplexity names four vendors and cites an analyst comparison, two vendor blogs and a review site. A brand that is named but not cited learns that the answer relies on third-party pages it does not control.
Common misunderstandings
- A numbered citation shows that a source was used, not that the brand it mentions was recommended.
- Blocking PerplexityBot is not a training opt-out, because Perplexity says that crawler is not used for training; it limits whether your pages can be surfaced in Perplexity's results.
- Answers change with the model in use and the sources available at that moment.