How does Kimi fetch web pages?
Kimi's official crawler policy page lists three crawlers with different jobs (opened and checked 2026-11-06; the page has no date):
| Crawler | Purpose (as stated officially) | If you block it |
|---|---|---|
| Kimi-SearchBot | Analyzes pages for relevance and builds the index for Kimi's search features | Your pages will not appear in Kimi search results |
| Kimi-User | User-initiated actions, such as summarizing an article or answering a question that needs live retrieval | Kimi warns it may reduce visibility in user-directed searches; robots.txt may not apply directly |
| KimiBot | May be used to train Kimi's foundation models | Signals your content should not be used for training |
So to appear in Kimi's search, allow Kimi-SearchBot and Kimi-User; blocking KimiBot only affects training. How to write it is in How do you write robots.txt for AI crawlers?; the minimum is this:
User-agent: Kimi-SearchBot Allow: / User-agent: Kimi-User Allow: /
Kimi also says each subdomain is a separate origin and needs its own robots.txt, and that you should not block by IP, since that may stop Kimi from even reading your robots.txt.
How does Kimi's search choose sources?
Kimi's open-platform web search documentation (an API for developers) says the Pro version returns relevance-ranked page chunks, each with the page title, link and site information, plus an authority field indicating how authoritative the source is. It does not name the underlying search engine or explain how authority is computed. This is API documentation; whether the Kimi app does the same is not stated: unverified.
How does Doubao pick its sources?
We found no official description of where Doubao's app takes its references from: unverified. What we did find is media coverage: Volcano Engine launched a "Doubao Search" service on 2026-07-28 and, according to the reports, builds an authority tiering by site and creator and filters low-quality information. We did not open Volcano Engine's official source, so treat this as a lead, not a fact.
What should a Chinese content site do?
Below, what is official and what is inference are kept apart.
| Approach | Why |
|---|---|
| Allow Kimi-SearchBot and Kimi-User | Official: Kimi's crawler policy page |
| Put body text, price, features and a one-line description in the HTML, not behind scripts | Inference: Kimi has a fetch interface that returns Markdown; whether it runs scripts is not stated |
| Give the Chinese page its own URL and a Chinese title instead of relying on browser translation | Inference: multilingual handling is not stated officially; see the multilingual guide for the general approach |
| Submit to both Baidu and Bing | Inference: whether Doubao's or Kimi's search relies on them is not stated; being findable in mainstream search is the general practice |
| Have honest descriptions of your product on Chinese communities, Zhihu and WeChat accounts | Inference: the "source authority tiering" comes from media reports: unverified |
For multilingual URLs and hreflang see How do you write hreflang on a multilingual site? How does AI search treat it?; for Bing indexing see How to set up Bing Webmaster Tools: import, sitemap and IndexNow step by step. DeepSeek is covered in How do I get DeepSeek to recommend my product?.
How do I know whether I am being recommended?
In Doubao and Kimi, ask the questions your buyers would ask, with web search on, three times each. Note whether you are mentioned and which reference pages are listed. Look at how many of the three runs include you, not at any single answer.
FAQ
What are Kimi's crawlers called?
Kimi lists three: Kimi-SearchBot (search index), Kimi-User (user-initiated fetches) and KimiBot (possibly training). Checked 2026-11-06.
Does blocking KimiBot make me invisible to Kimi?
Kimi says blocking KimiBot signals your content should not be used for training; blocking Kimi-SearchBot is what keeps you out of search results.
Which sites does Doubao cite?
No official explanation was found: unverified. The media-reported site tiering was not checked against an original source either.
Do I need to be indexed by Baidu?
This is inference: whether Doubao's or Kimi's search depends on Baidu is not stated. Being findable in both Baidu and Bing is the general practice.
Can I be guaranteed a recommendation?
No. Answers differ each time and the ranking is not published.
Sources
- Kimi · Kimi Crawlers （Checked 2026-11-06; the page has no date）
- Kimi · 开放平台 · 联网搜索 （Checked 2026-11-06）
- kagentx evaluations: the data on the case study pages; each example gives its date.