Chat With Global News in n8n: Webz.io + GPT-4.1-mini

Ask global news questions inside n8n—backed by sources

This n8n workflow lets you chat with global news using Webz.io News Search and an OpenAI GPT-4.1-mini agent. Get answers with article URLs for each claim, plus ongoing context for follow-up questions—all in a streaming chat experience.

What this workflow does

• Receives your question from the n8n chat panel and starts a streaming response.
• Uses a GPT-4.1-mini agent to interpret what you’re asking and automatically derive search filters such as language, country, date range, sentiment, category, domain, ticker, and entities.
• Queries Webz.io global news via the Webz.io News Search MCP endpoint using those filters.
• If the initial query returns no results, the agent refines the search and retries.
• Maintains conversation memory so you can ask follow-ups (e.g., “compare to last week,” “focus on another region,” “only from a specific publisher”) without repeating constraints.
• Responds with answers grounded in source URLs, or clearly states when no relevant results are found.

Use cases

• Build an automation-first news brief for your SaaS team: “What happened with [topic] in [region] this week?”
• Monitor market or company updates by using ticker/entity constraints and getting sourced summaries.
• Support product decisions with contextual follow-up questions in the same chat thread.

Technical details

• n8nn8n-nodes-langchainchat trigger for the chat entry point
• n8nn8n-nodes-langchainlm chat open ai configured with GPT-4.1-mini
• n8nn8n-nodes-langchainmcp client tool calling Webz.io News Search MCP (Bearer token credential)
• n8nn8n-nodes-langchainagent to orchestrate tool-calling and search filter derivation
• n8nn8n-nodes-langchainmemory buffer window for conversation context
• Streaming output and workflow UI elements (e.g., sticky note)

Setup (quick)

• Create a Webz.io API token and add it as a Bearer Auth credential for the Webz.io News Search MCP tool.
• Add an OpenAI API credential and ensure the chat model is set to GPT-4.1-mini (or replace with another tool-calling chat model).
• Open the workflow’s chat panel and start asking.