Give customers fast answers with a smart chat that searches the web and replies in plain language. Ideal for support teams and product communities that need current info without manual research. It helps agents and users get helpful results in one place.

Incoming messages start a chat session. An AI agent reads the message, checks short term memory for context, and uses GPT 4o to plan a response. When it needs fresh data, the agent calls Brave Search through MCP tools. The flow pulls the right tool, builds the query, runs the search, and returns the summary to the user. This short loop keeps answers relevant and reduces back and forth.

Setup is simple. You need an n8n instance, an OpenAI API key, and a Brave Search API key. Connect both keys in n8n, confirm the MCP tool list loads, and test a few questions in the chat interface. Expect quicker replies, lower handle time, and fewer escalations. Great for help widgets, status checks, and content questions that change often.

• Chat trigger starts a session whenever a new message arrives
• AI agent plans steps and decides when to call web search tools
• GPT 4o generates clear replies and interprets user intent
• Simple Memory keeps short term context so follow up questions make sense
• MCP Get Brave Tools lists available Brave Search capabilities
• MCP Execute Brave Search runs the selected tool with dynamic parameters
• Agent fills tool name and parameters automatically using AI signals
• All pieces connect inside n8n for a single place to manage chat logic

• Reduce manual web searches from 10 minutes to under 30 seconds per question
• Automate up to 70 percent of common info requests with AI guided search
• Cut first response time for chat by 60 percent using instant lookups
• Handle three times more concurrent chats by removing context switching
• Keep answers current by pulling results from live web data

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with OpenAI and Brave Search. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create an OpenAI credential. If unsure, double click the GPT 4o node, choose Credential to connect with, click Create new credential, and follow the on screen steps to paste your OpenAI API key from your OpenAI account API page.
4. Create a Brave Search API key from the Brave Search API page. In n8n, double click each MCP node, choose Credential to connect with, click Create new credential for MCP Client, and follow the on screen steps. Add your Brave API key where requested by your MCP setup.
5. Open the MCP Get Brave Tools node and run it once to confirm a list of tools is returned. If it fails, check the API key and that MCP Client can reach the Brave tool.
6. Open the MCP Execute Brave Search node and keep the tool selection fields as configured so the agent can set toolName and parameters dynamically. Do a test run with a sample query to verify results come back.
7. Check the Simple Memory node and set the window size to match your chat needs. A small window keeps the chat fast while holding recent context.
8. Review the AI Agent node prompt and role. Add any brand tone or support policies you want the model to follow.
9. Start the workflow and open the n8n chat interface. Ask a question that needs current info. Confirm the reply includes recent details drawn from the web.
10. If you see 401 errors, recheck API keys. If no tool is selected, confirm the MCP Get Brave Tools node returns tools. For slow replies, lower the memory window or simplify the prompt and retry.