n8n Multi-Agent Cold Email Campaign Generator: O3 Director + GPT-4.1 Specialists

Build a chat-triggered multi-agent workflow in n8n: an O3 Director Agent orchestrates strategy while six GPT-4.1-mini Specialist Agents run research, copywriting, personalization, sequencing, deliverability, and analytics in parallel, then integrate into a complete cold email campaign plan.

Who

Sales and marketing operators who want to quickly build an AI-powered cold email campaign in n8n and reduce OpenAI costs via agent specialization

You get

Receive a complete cold email campaign plan—including prospect research, email copy, personalization variables, follow-up sequence, deliverability tips, and analytics—by sending a natural-language request via the n8n chat trigger

You need

n8n · chat-trigger · ai-agent · agent-tool · openai

Steps

1. Add the Chat Trigger node

   Add a "When chat message received" node (type @n8n/n8n-nodes-langchain.chatTrigger, version 1.1) to the workflow canvas. The default webhookId is outreach-webhook-id; no extra parameters are required to receive user requests through the n8n chat interface. Once saved, the workflow can be triggered from the built-in chat or external messages.

2. Configure the Outreach Director Agent (O3)

   Add an "Outreach Director Agent" node (type @n8n/n8n-nodes-langchain.agent, version 2.1). Connect its language model to the "OpenAI Chat Model Director" node with model set to o3. The Director parses user requests, formulates campaign strategy, and delegates tasks to downstream Specialist Agent Tools.

3. Connect the Think tool for the Director

   Add a "Think" node (type @n8n/n8n-nodes-langchain.toolThink, version 1.1). Connect its ai_tool output to the Outreach Director Agent's tool input. The Director can invoke this tool for strategic planning before delegating tasks.

4. Add and configure the six Specialist Agent Tools

   Add the following six Agent Tool nodes (type @n8n/n8n-nodes-langchain.agentTool, version 2.2), each connecting its ai_tool output to the Outreach Director Agent: 1. Prospect Research Specialist — model via OpenAI Chat Model1 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in prospect research, company analysis, pain point identification, and personalization insights" 2. Cold Email Copywriter — model via OpenAI Chat Model2 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in persuasive email copy, subject lines, hooks, and conversion-focused messaging" 3. Personalization Specialist — model via OpenAI Chat Model3 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in email personalization, custom variables, dynamic content, and individual prospect customization" 4. Email Sequence Strategist — model via OpenAI Chat Model4 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in email sequence design, follow-up timing, nurture campaigns, and multi-touch outreach strategies" 5. Email Deliverability Expert — model via OpenAI Chat Model5 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in email deliverability, spam avoidance, domain reputation, and inbox placement optimization" 6. Outreach Analytics Specialist — model via OpenAI Chat Model6 (gpt-4.1-mini), toolDescription: "call this AI Agent that specializes in email performance tracking, A/B testing, response rate optimization, and campaign analytics" Each Specialist node's text parameter uses the same expression:

   Execution prompt · Source text。Text parameter field of the six Specialist Agent Tool nodes; receives the user message dynamically from the Director at runtime

   Copy prompt

   ={{ $fromAI('Prompt__User_Message_', ``, 'string') }}

5. Configure language model nodes and connections

   Add seven OpenAI Chat Model nodes (type @n8n/n8n-nodes-langchain.lmChatOpenAi, version 1.2): - OpenAI Chat Model Director: model set to o3, ai_languageModel output connected to Outreach Director Agent - OpenAI Chat Model1–6: model set to gpt-4.1-mini, each connected to the corresponding Specialist Agent Tool's ai_languageModel input Verify all connections: Chat Trigger → Director; Think → Director (tool); six Specialists → Director (tool); seven Models → respective Agents (language model).

You are setting up the n8n template 6909 multi-agent cold email campaign generator. First confirm: 1) your n8n instance is accessible and an OpenAI API Key credential is configured; 2) you understand the分工: Director Agent uses o3, six Specialist Agent Tools use gpt-4.1-mini; 3) Specialist nodes use `={{ $fromAI('Prompt__User_Message_', ``, 'string') }}` in the text field to dynamically receive user input from the Director, so no fixed prompt is pre-written. If you need to customize a Specialist's system behavior, add it in the agentTool options or an upstream node rather than changing this dynamic expression.