AI assistant fetcher
Perplexity-User
Fetchers that load a page live because a person asked an AI assistant about it.
Last updated
At a glance
| Category | AI assistant fetcher |
|---|---|
| Operator | Perplexity |
| User agent match | contains "perplexity-user" |
| Executes JavaScript | No |
| Respects robots.txt | Generally yes |
| Identity verifiable | Yes, against published IP ranges |
| IVT class | GIVT (general invalid traffic: declared automation) |
What Perplexity-User is
Perplexity-User fetches a page at the moment a person asks an AI assistant about it. It is closer to a proxy for a human reader than to a crawler on a schedule: each visit is one page loaded because one person wanted its content right then. It does not execute JavaScript.
Perplexity documents this bot, and the primary source beats any directory, ours included: official Perplexity-User documentation.
Can the name be trusted?
The operator publishes the IP ranges Perplexity-User calls from, and TrueVisit checks every visit claiming this name against them. A claim arriving from an address outside the published ranges fails verification and is reported as exactly that, so an impostor borrowing the name cannot borrow the trust.
Should you block it?
Think twice before blocking Perplexity-User: each of its requests is a person actively asking an assistant about your page. Blocking it makes your content unquotable in that conversation, which is the opposite of what most publishers want. Block it only if you have decided you want nothing to do with AI assistants at all.
User-agent: Perplexity-User
Disallow: /For enforcement rather than a polite request, the Cloudflare blocking guide generates copy-paste WAF rules from this same signature list, with guardrails that keep verified search engines safe.
How TrueVisit reports it
Visits from Perplexity-User appear in your bot report under this exact name, never as an anonymous percentage, and every verdict stores the reason it was reached. Because this client never executes JavaScript, a browser-only tracker cannot see it at all; it is visible through server-side collection, which is how the visits reach your report.