• ENTITIES
  • TOPICS
  • RELATIONS
  • HUMAN

How to extract entities and topics from URLs with your AI

Connect TextRazor to your AI client once. After that, your agent can parse any web page you send it.

Ask AI about this page

Short answer

How can I get entities and topics from a URL using AI?

Give your AI a URL and it uses TextRazor to pull out specific entities, topics, and the relations between them. You get structured data instead of just raw text. This breakdown shows what you'll get.

Data outcomes

What your AI extracts

The specific pieces of information your agent pulls from a link.

ENTITIES

Named entities identified

The AI finds people, places, and organizations. You get a clean list of the specific subjects mentioned on the page.

TOPICS

Core topics found

Your agent identifies the main themes. This lets you understand the subject matter without reading the whole page.

RELATIONS

Connections mapped

The tool identifies how different entities relate to each other. You see the links between the people or things in the text.

HUMAN

Manual review needed

The AI provides the structured data, but you decide which entities are actually relevant to your work.

The workflow

What your AI does when a URL arrives.

The AI handles the heavy lifting of reading and parsing the web content.

  1. Fetch text

    Your agent accesses the URL to read the text content.

    analyze_text

  2. Identify entities

    The AI scans the text for specific people, locations, and organizations.

    analyze_text

  3. Extract topics

    The agent identifies the main subjects and themes within the text.

    analyze_text

  4. Find relations

    The AI maps how the identified entities connect to one another.

    analyze_text

Try it

Copy these to start.

Paste these into your chat to begin extracting data.

Starting points

These are opening lines. Replace the bracketed text with your specific URL or target.

TextRazor Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_vecqfh8K0a3E1sxbBP1vh9Y5BRM4TTm2vrIW6dfx/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — TextRazor capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "textrazor-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_vecqfh8K0a3E1sxbBP1vh9Y5BRM4TTm2vrIW6dfx/mcp"
    }
  }
}
Copy into chat04

  • Extract all entities and topics from this URL: [URL]

  • What are the main topics and people mentioned at [URL]?

  • Find the relations between entities on this page: [URL]

  • List the organizations mentioned in this article: [URL]

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect TextRazor once, then ask.

Connect the MCP to your client, and your login stays encrypted on our side. You can find the full walkthrough on the Connector page.

Connect TextRazor to your AI

FAQ

How this task behaves

  • 01

    Can the AI edit the webpage?

    No. The AI only reads the content to extract data; it cannot change the website.

  • 02

    Does the AI save these entities to a file?

    No. The AI provides the data in your chat. It won't save it to a file or database unless you manually copy it.

  • 03

    Can I use this for a PDF?

    No. This task is built for URLs. You would need to provide the text directly if you want to analyze a PDF.

  • 04

    What if the website is behind a login?

    No. If the AI cannot access the URL because of a paywall or login, it cannot extract any data.

  • 05

    Does it work on any website?

    It works on any URL that is publicly accessible and readable by your agent.

  • More questions about TextRazor? The Connector page answers them. See everything the TextRazor Connector can do

Connect TextRazor to Claude, Cursor, ChatGPT & more