This guide walks through the same Lab 2 learners complete in our Introduction to Agentic AI for Business Users course. In this course, learners move beyond prompts to understand how AI agents are designed and orchestrated. Using no-code, hands-on labs in n8n, the course covers core agent components, real business use cases, and the ethical and security considerations required for responsible deployment.
This lab appears early in the Agentic AI for Business Users course and serves as a foundational building block for everything that follows. Before learners can design autonomous agents, they need to know how AI works inside a workflow. This includes how prompts are sent, how responses are created, and how behavior is changed through design choices. In this lab, learners build a simple AI‑enabled workflow using n8n, and will:
- Connect n8n securely to an OpenAI model
- Build a simple AI workflow using visual nodes
- Understand how System and User prompts shape AI behavior
- Test and refine AI responses in real-time
Prerequisites
- Access to n8n Cloud (credentials provided in class)
- An OpenAI API key with billing enabled
- A modern web browser (Chrome, Firefox, or Edge)
- A stable internet connection
Step 1: Launch n8n
In our instructor-led class, we provide a fully configured lab environment so learners can focus on understanding workflows and AI concepts and not worry about the setup. If you’re following this guide outside of class, you can still complete the lab by using n8n Cloud directly by following these steps:
- Open a browser and go to: https://n8n.io
- Select Start for free (or sign in if you already have an account)
- Create an n8n Cloud account and log in
- Once logged in, you should land on the n8n dashboard, where you can create and manage workflows
Understanding AI in n8n
n8n provides dedicated nodes for interacting with AI models. In this lab, we will focus on the OpenAI Messaging node, which allows you to send text prompts directly to models like GPT-4o.
Key Concepts:
- Nodes: Building blocks of n8n workflows that perform specific tasks (e.g., triggers, actions).
- Triggers: Nodes that start workflows based on events (e.g., manual trigger, chat trigger).
- Actions: Nodes that perform operations in response to triggers (e.g., sending a message, making an API call).
- Credentials: Securely stored API keys that allow n8n to talk to external services like OpenAI.
- Roles: The context in which a message is sent (e.g., 'System' defines behavior, 'User' provides input).
Configuring OpenAI Credentials
Before using AI models, you need to configure OpenAI API credentials to access the LLM in n8n. This means that we have to store your OpenAI API key securely within n8n for use in all future workflows.
Open the Credentials section in n8n:
- From the HOME n8n dashboard, click Credentials tab in the central menu.
- Click Create Credential button in the top right.
- In the search box, type OpenAI and select OpenAI.
- Then click Continue.
- Enter your OpenAI API key by pasting it from the provided lab credentials under the field labeled API Key (in OpenAI keys document from student portal*)
- Click Save.
- When you press Save, n8n will validate the key and should indicate success.
* If you are following this guide on your own, you will need to create and use your own OpenAI API key.
- Open a browser and go to: https://platform.openai.com/api-keys
- Sign in or create an OpenAI account
- Click Create new secret key
- Copy the key (you won’t be able to see it again)
Setting Up Your Workflow
- From the n8n dashboard, click Create Workflow.
- Name it Lab 2 - Enabling AI.
- Save the workflow. Now let’s build a simple workflow that uses an AI model.
Adding a Trigger Node
Triggers allow us to start workflows based on events. We can make various types of triggers:
- Manual Trigger: Start the workflow manually.
- Chat Trigger: Start the workflow when a chat message is received (covered in later labs).
- Action Trigger: Start the workflow based on an action in another app (e.g., new email).
For this lab, we’ll use a Manual Trigger.
- Return to your workflow canvas.
- Click + to open Node Selection Menu and search or scroll to find option to "Trigger Manually."
- Select Trigger Manually option.
Important: The Manual Trigger starts your workflow when you click Test Workflow.
Adding an AI Model
- Click the + that appears to the right of the Manual Trigger node.
- Search for OpenAI and select OpenAI (Tip: Make sure to select the OpenAI node, not other similarly named nodes.)
- Review the OpenAI options menu, which contains various operations like: Assistant Actions, Text Actions, Image Actions, and more. These are options for different ways to interact with OpenAI models.
- For now, select the Message a Model option.
Note: The Message a Model operation sends a direct request to the AI and gets a response. This is different from an "Agent" (which we will cover in Lab 3) that can autonomously use tools and loop.
You should see the following configuration options:
- Select your OpenAI credential from the dropdown (if prompted).
- Set Resource to Text. This tells the node we are sending a text-based prompt, rather than an image or audio file.
- Set Model to a cost-effective option (for example, gpt-4o-mini or gpt-5-nano).
- Enter a prompt in Prompt like: Hello, help me develop n8n workflows with openai!
- Leave Role as user.
- Run the workflow by clicking Execute Step at the top of the configuration panel.
- Press kEsc again to see the main canvas board with our current workflow.
Tip: If the response reaches an error, there may be an issue with your OpenAI credentials or model selection. Double-check those settings and try again.
CheckPoint: You should see a workflow like below in the main canvas view.
Modifying Inputs to the AI Model
We can modify the prompt and other settings to change the interaction with the AI model. Let’s try that now.
- Double-click the OpenAI Message a Model node.
- We previously used the User Role, but now we can try out the System role. In the dropdown labeled Role, set the value to system. Tip The System role is used to define the AI’s behavior, personality, and constraints. It sets the "rules" for the conversation.
- In the Prompt field, change the message to: You are a friendly assistant that provides interesting facts about technology and automation.
- Add a new message using the Add Message button in this configuration.
- For the new message, set Role to User and Message to: (Tip: The User role represents the input from the human or the trigger. This is the actual question or task you want the AI to perform.)
- Click Execute step to test the change.
- Observe how the AI now adopts the persona defined in the System prompt while answering the User prompt.
YOUR TURN: Try changing the prompts and roles further to see how the AI’s responses vary. Experiment with different instructions in the System role and questions in the User role.
Completion Checklist
- Added and configured OpenAI credentials
- Built a workflow with Manual Trigger and AI node
- Successfully tested the workflow
- Created first LLM workflow
About Ascendient Learning
Ascendient Learning, a part of Accenture LearnVantage, offers live, hands-on AI and Agentic AI training. Our courses focus on learning practical skills that can be used in your daily work immediately. Training is delivered live online or at your site by experienced practitioners, combining clear instruction, guided labs, and an emphasis on responsible AI practices.
Whether you’re enabling business users, technical teams, or cross-functional audiences, learners leave with confidence in how AI works and how to apply it in their roles. Contact Ascendient Learning to learn how we customize training to help organizations develop practical AI capability across their workforce.
