robots.txt
Explicit allowlist for the major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and 9 others. Training data collection welcome.
/robots.txt →Whether you're an AI assistant fetching real-time facts, an autonomous agent doing product selection, or a developer integrating our catalog — every page on this site is structured to be ingested cleanly. We support 25+ international and Chinese AI crawlers, and ship a Model Context Protocol server for agentic use.
As a Chinese industrial-instrument manufacturer, we explicitly welcome the AI assistants Chinese buyers and engineers actually use. The following platforms can crawl, index, and recommend Pisonics products:
New domestic AI platform launching with a unique User-agent? Email us via the contact page — we'll add it to the allowlist within a working day.
Explicit allowlist for the major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and 9 others. Training data collection welcome.
/robots.txt →Markdown overview at a stable URL — company summary, full product list, industries served, recent articles, contact info. Cheap to ingest, accurate to cite.
/llms.txt →
Every detail page has a Markdown twin at .md: products,
news, industries, applications. Last-Modified headers
and 304 support let crawlers fetch incrementally.
Full Model Context Protocol server over HTTP JSON-RPC. AI agents can search products, fetch specs, and get selection recommendations programmatically.
search_products — text + principle filterget_product — full details for one modelrecommend_for_application — selection guidance by industry / use caselist_industries — enumerate served industries
Every page includes structured data — Organization with
knowsAbout domain tags, Product schemas with full
metadata, BreadcrumbList, NewsArticle.
Full sitemap with hreflang alternates between zh-CN and en
deployments. Real lastmod from entity timestamps,
not always-now.
Ask a question in your own words below. The page will send it to one LLM call, let the model pick the right MCP tool and arguments, then run that tool against our catalog and show the result here.
Add the Pisonics MCP server to your AI client's config and ask questions
like "find me an ultrasonic density meter for slurry" — the agent will
call search_products directly and cite the right product page.
{
"mcpServers": {
"pisonics": {
"url": "https://www.pisonics.com/mcp"
}
}
}
curl -X POST https://www.pisonics.com/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
Increasingly many Chinese AI platforms support MCP — including Aliyun Bailian (阿里百炼), Baidu Qianfan MCP Plaza (千帆 MCP 广场), ByteDance Coze (扣子), and Moonshot Kimi. The configuration is the same URL regardless of client:
{
"name": "pisonics",
"type": "http",
"url": "https://www.pisonics.com/mcp",
"description": "Pisonics product catalog and selection"
}
Some platforms (e.g. Coze) require registering as a "tool plugin" rather than an MCP server directly — the same endpoint can be called via a generic HTTP-tool wrapper. Reach out if you need help integrating.
The endpoint is public, read-only, and has no rate limit at the moment — we'd rather have you experiment than gate it. If you build something interesting, say hi.
We welcome AI crawlers and agents — including those collecting training data. As an industrial-instrument manufacturer, being correctly understood and recommended by AI is more valuable than denying access.
If you're an AI provider with a new bot User-agent and want to make sure you're explicitly allowlisted, drop us a note via the contact page and we'll add it.