Perplexity-User
Perplexity-User is the fetcher Perplexity sends to read a page when someone asks it a question, as opposed to crawling the web ahead of time.
Perplexity-User is the name of the fetcher Perplexity uses when a person's question needs a page read right now. Someone asks Perplexity something, Perplexity decides it should look at a page, and Perplexity-User goes and gets it. The page can then be linked in the answer.
It matters because it behaves differently from the bots you might already be managing in robots.txt. Perplexity says in its crawler documentation that because a user asked for the fetch, Perplexity-User generally ignores robots.txt rules.
How it differs from PerplexityBot
Perplexity runs two bots, and they do different jobs.
PerplexityBot crawls the web in advance so Perplexity can find your pages and link them in its search results. Perplexity-User only visits when a question calls for it, one page at a time. Perplexity says neither one collects content to train AI foundation models.
So a rule in robots.txt that names PerplexityBot controls the first. It may not stop the second.
An example
Say Fernhill Coffee blocks PerplexityBot in robots.txt. Someone asks Perplexity: "What does Fernhill Coffee's subscription cost? Check their pricing page."
PerplexityBot can't crawl the site, so the pricing page may never have been picked up ahead of time. But if Perplexity fetches it for that question, it does so as Perplexity-User, and Fernhill's robots.txt rule doesn't apply to that visit.
If you want to keep Perplexity-User out, robots.txt alone won't do it. You'd need a rule at your host or CDN, the same way you'd turn away any other visitor.
Perplexity-User in AnswerOrbit
The Audit tab in Website lists each AI bot with its purpose, and marks the ones that fetch for a user. If you connect your server logs, the Bot visits tab shows which of these bots read your pages, and we check each request against what the company publishes before we count it. See Website and AI traffic.