A public endpoint to query Busony in natural language
NLWeb turns our website into a database AI agents can query. No HTML scraping: a natural-language question in, structured Schema.org results out.
An open protocol, not a chatbot
NLWeb (Microsoft, open source) turns a website into an endpoint AI agents can query — not humans. An agent sends a free-text question and gets back a list of structured results (Schema.org) with a relevance score, without ever having to parse our pages' HTML. That's what lets an agent compare Busony to other providers on real criteria, not an impression pulled from a poorly structured page.
Two ways to query Busony
GET request, streamed as Server-Sent Events (text/event-stream), one JSON object per line.
GET /api/ask?query=...Standard MCP server (initialize, tools/list, tools/call) — for Claude and compatible agents.
POST /api/mcpExample
https://busony.com/api/ask?query=services+for+exporting+SMB&site=busonyResponse (excerpt)
Content-Type: text/event-stream
data: {"message_type": "result", "content": [{
"@type": "Item",
"url": "https://busony.com/en/exportik",
"name": "Exportik",
"score": 94,
"schema_object": {"@context": "https://schema.org", "@type": "Service", ...}
}]}Note : Public access, no API key required. Rate limited to 10 requests/minute per IP (HTTP 429 beyond that).
The queryable catalog today
services
blog articles
entries total
Automatically re-indexed on every notable content change — the catalog reflects the pages and articles actually published on busony.com.
How an agent finds this endpoint
robots.txtA schemamap directive pointing to our Schema Map — an emerging convention from the NLWeb ecosystem.
schemamap.xmlReferences our Schema.org feed in JSON Lines format.
llms.txtEndpoint documentation for AI agents and assistants.
This pageA plain HTML link, discoverable by Google, ChatGPT and Perplexity like any other content.
Want the same endpoint on your site?
Busony deploys NLWeb, WebMCP and MCP for its clients — based on what we built and measured for busony.com itself.