Getting set up

1. Getting set up

Before we start building, let me walk you through a few things that will help you work smoothly through this course.

2. Your n8n workspace

We've pre-configured an n8n instance for you in the exercises. You'll see the exercise steps on the left side and n8n on the right side. Everything is ready to go. No installation or setup required, just open the exercise and start building. OpenAI credentials are already connected in your instance, so you don't need to bring your own API key. The only account you'll create yourself is a free Apify account in Chapter 3, which you'll use to pull business listings from Google Maps. We'll walk you through the setup step by step when we get there. One important note: if you leave your session and return after about 30 minutes, your work may be lost. But don't worry! We've uploaded checkpoint files you can import at any time. Just create a new workflow, click Import from File, go to Desktop, then Resources, and you'll find JSON files organized by chapter and exercise number. These checkpoints may come in handy if you simply need to check the solution of any exercise. Now, let me walk you through what you'll build in the first three exercises. You'll create a new workflow and name it "Marketing Multi-Agent System." You'll add a Manual Trigger node, then a Set node called Campaign Brief. This node holds four fields: business context, objective, target audience, and budget. It's the shared brief that all three specialist agents will read. Then, you'll build three AI Agent nodes in parallel branches: a Brand Agent, a Content Agent, and a Paid Ads Agent. Each agent gets its own OpenAI Chat Model and its own prompts—a user message that references the campaign brief fields, and a system message that gives the agent its specialist role. Because they branch from the same node, each one works independently, seeing only the original brief. Then, you'll add a Merge node to collect the three outputs, followed by a Code node that reshapes them into a single combined object. This is what allows the final agent to run exactly once. The last piece is the CMO Summary Agent. It reads the three strategies and produces a single executive brief with a 90-day action plan. You'll also add a Structured Output Parser, which forces the agent to return a consistent JSON structure with an executive summary, top priorities, and day 30, 60, and 90 milestones. By the end of these exercises, you'll have a fully automated multi-agent marketing department: three specialists working in parallel branches, and one CMO synthesizing everything into a structured, actionable brief.

3. Let's get started!

That's everything you need to know for now. Time to start building!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.