How to Get Your Business Mentioned by Claude: Anthropic’s Three Crawlers and the Brave Connection

To get your business mentioned by Claude, allow Anthropic’s search crawlers, Claude-SearchBot and Claude-User, in robots.txt and at your firewall; make sure your pages are also crawlable by Googlebot and appear in Brave Search, which independent tests found behind Claude’s web search; and get mentioned on the third-party pages that answer your customers’ questions. When Claude doesn’t search, it answers from training, where only broad, consistent mentions across the web help.

Anthropic’s three crawlers, and what blocking each does

Anthropic says its bots “respect ‘do not crawl’ signals by honoring industry standard directives in robots.txt”, and they support Crawl-delay. That makes Claude-User unusual: OpenAI’s ChatGPT-User, Perplexity-User and Google’s user-triggered fetchers may ignore robots.txt, but a Disallow for Claude-User does stop Claude fetching your page for a user. If you want to be cited, allow it. Anthropic also notes that robots.txt is per subdomain, and that blocking its IP addresses instead “may not work correctly”.

User-agent: Claude-SearchBot
Allow: /

User-agent: Claude-User
Allow: /

# Training is a separate choice:
User-agent: ClaudeBot
Disallow: /

Copy any rules from your User-agent: * group into these groups too, because a crawler follows only the most specific group that names it. The AI robots.txt generator writes the whole file for search-yes, training-no.

The Brave connection

Anthropic’s documentation for its web search tool doesn’t name a search provider. The evidence comes from outside. On 21 March 2025, Simon Willison reported that Brave Search had been added to Anthropic’s subprocessor list on 19 March, that Claude’s web search tool definition included a parameter set named “BraveSearchParams”, and that the ten pages Claude cited for a test query matched Brave’s results for the same query exactly. TechCrunch reported the same day that Anthropic appeared to be using Brave. Anthropic hasn’t announced a change since, but it hasn’t confirmed the arrangement in its docs either.

If Claude’s searches do run through Brave, two practical things follow:

• Check you’re in Brave Search. Search search.brave.com for your business name and your customers’ questions. If you don’t appear for your own name, that’s worth fixing before anything else.
• Don’t block Googlebot from anything you want Claude to find. Brave says its crawler doesn’t advertise a separate user agent, and that “if a domain or page is not crawlable by Googlebot, then Brave Search’s bot will not crawl it either.”

When Claude doesn’t search

Claude answers many questions without searching, from what it learned in training. You can’t submit facts to that. What shapes it over time is how widely and consistently your business is described on the open web: in Ahrefs’ study of 75,000 brands, branded web mentions correlated with ChatGPT visibility far more strongly than backlinks did, and the same logic applies to any model trained on the web. Coverage, directories, reviews and lists in your field are the work; see how to help AI understand your business.

Pages Claude can quote

• Key facts in the HTML text, not loaded by JavaScript.
• The answer in the first lines, with specifics: what, for whom, where, how much.
• A real, visible date on pages whose facts change.
• A clear About page that says who is behind the business.

Check and measure

Ask Claude your customers’ questions with “search the web” added, three times each, and record the cited pages; then ask without it to see what it knows from training. Claude often differs from the others: in Custom Legal Marketing’s 3,200-prompt study of law firms, Claude, ChatGPT and Google AI Mode rarely recommended the same firms. Our AI visibility checker, rolling out engine by engine, asks Claude with live web search and shows the sites it cites, and the [free audit](/) checks access for all three Anthropic crawlers. For the other engines, see Perplexity, Gemini and Copilot.