How to use Gemini 4 Argon in n8n for your LinkedIn work

What is Gemini 4 Argon?

Google's release notes for the Gemini app call it its newest frontier model and say it is rolling out to Google AI Pro and Ultra subscribers there Google: Gemini Apps release updates and improvements. In its words Google: Gemini Apps release updates and improvements:

"4 Argon is our newest frontier model, built to take your most complex tasks from start to finish."

Google: Gemini Apps release updates and improvements

This page is about another use of it: as the model inside an n8n agent that works on your LinkedIn.

Can an n8n agent run on Gemini 4 Argon?

Yes, by n8n's own post, which says this n8n blog: Build n8n workflows with Gemini 4 Argon: MCP, the Assistant, and Agents:

"Gemini 4 Argon is available as a model for your Agents and in the standalone Google Gemini node from launch."

n8n blog: Build n8n workflows with Gemini 4 Argon: MCP, the Assistant, and Agents

The same post says an n8n agent is a model handed a goal and the tools it is allowed to use, any node, API or sub-workflow, and that it decides which tool to call next n8n blog: Build n8n workflows with Gemini 4 Argon: MCP, the Assistant, and Agents. On n8n Cloud, it says, you can use Argon through Gateway credits without a separate provider account n8n blog: Build n8n workflows with Gemini 4 Argon: MCP, the Assistant, and Agents.

How do I give that agent my LinkedIn?

1. Add an AI Agent node, and choose Gemini 4 Argon as its model.
2. Add an MCP Client Tool node under the agent, with your LinkedIn address and key, as the n8n page shows.
3. Give the agent only the actions this workflow needs. A workflow that reads needs no action that sends.

That last step is the hard line n8n's post talks about: the agent can only reach the tools you plug in n8n blog: Build n8n workflows with Gemini 4 Argon: MCP, the Assistant, and Agents.

What to have the agent do

Write the agent's task the way you would say it to a person:

• "Every weekday at 9, list who on LinkedIn is waiting on a reply from me, read each conversation, and post a short summary with a suggested answer to my Slack. Send nothing on LinkedIn."
• "When a row is added to the sheet, find that person on LinkedIn and read their profile and recent posts, then write three lines about them back into the row."

• who_is_waiting_on_meList the people whose last message you have not answered.
• get_conversationRead your whole conversation with one person.
• search_peopleSearch people on ordinary LinkedIn by plain words, a place name and connection degree.
• get_profileRead one person's profile in short, with their recent posts and their company.
• list_posts_by_authorList the posts one person or one company page published, newest first.

Both read only. Sending stays with you: you read the suggestion and answer from LinkedIn or from your AI app when you choose.

Will a scheduled workflow stay connected?

On the free plan LinkedIn is disconnected after a while without use, so a workflow that runs the next morning finds it disconnected. The plan that keeps LinkedIn connected for scheduled work is on the pricing. If you would rather not build the workflow at all, the LinkedIn AI agent finds who to talk to next and asks you before it sends.

Where to set it up

The address, the key and the MCP Client Tool node are on the n8n page. What each action does, and its daily limit, is on the actions page.