How to create any n8n workflow using ChatGPT

Key Concepts

• NA10 workflow automation
• AI agent creation using ChatGPT
• JSON workflow files
• Bug report collection and management
• OpenAI models (GPT-3, GPT-4)
• Google Sheets integration
• Gmail integration
• Hostinger VPS deployment
• Prompt engineering
• AI-powered bug triage

NA10 Workflow Creation with ChatGPT

Project Setup in ChatGPT

1. New Project: Create a new project in ChatGPT specifically for NA10 workflow building.
2. System Prompt: Define a detailed system prompt that instructs ChatGPT to act as an AI agent designed to create JSON workflow files for NA10. The prompt should specify how the JSON files for NA10 modules should be structured.
   • Example: "You are an AI agent designed to create any JSON workflow files..."
3. Example Files: Add example JSON files of existing NA10 workflows to the project. These files serve as references for ChatGPT to understand the desired output format. Four example files are used in the video.
   • These files and prompts are available in the "New Society" (link below the video).

Building a Bug Form Collection Agent

1. Agent Description: The goal is to create an agent that collects bug reports via a form, sorts them by type, and then uses specialized AI agents to check a Google Sheet for existing reports. If the bug is new, it's appended to the sheet and the team is notified via email.
2. Prompting ChatGPT: Send a prompt to ChatGPT describing the desired bug reporting system, including the form input, bug type selection, and NA10 workflow outline.
   • Example: "You're creating a system where someone can report a bug through a form..."
3. Model Selection: Use the GPT-3 model (specifically "03" in the video, but "4.1" is recommended later) for advanced reasoning capabilities. The paid version of ChatGPT is recommended for access to these models.

Workflow Generation and Import

1. JSON Generation: ChatGPT generates a JSON workflow file based on the provided prompt, example files, and system prompt. This process may take some time (e.g., 1 minute 20 seconds in the video) due to the reasoning involved.
2. Workflow Download: Download the generated JSON file from ChatGPT.
3. NA10 Import: In NA10, create a new workflow and import the downloaded JSON file.
4. Initial Error Handling: The imported workflow may contain errors that need to be addressed.

Workflow Configuration and Customization

1. Visual Organization: Reorganize the nodes in the NA10 workflow for better readability and understanding.
2. Bug Report Form Configuration:
   • Set the form title (e.g., "Report Your Bug").
   • Add a description (e.g., "Take two minutes to quickly report the bug you found").
   • Add form elements (questions):
     • "What kind of bug is it?" (drop-down list with options like UI bug, notes related, tasks related, agent related, other).
     • "Describe the bug in detail" (text field for detailed description and replication steps).
   • Mark fields as required.
3. Switch Node Configuration: Configure the switch node to route bug reports to different AI agents based on the selected bug type.
   • Each option in the drop-down list should have a corresponding route in the switch node.
   • The "Routing Rules" should match the exact keywords used in the form options (e.g., "UI bug," "notes related").
4. AI Agent Setup:
   • Add an OpenAI model to each AI agent node.
   • Connect an OpenAI credential (API key). Instructions are provided to obtain an API key from platform.openai.com.
   • Select the GPT-4.1 model (recommended for its cost-effectiveness and reliability).
   • Define the user prompt for the AI agent, including the bug type and user-provided description.
     • Example: "Bug type: {{ $json["What kind of bug is it?"] }}\nUser bug description: {{ $json["Describe the bug"] }}"

Tool Configuration (Google Sheets and Gmail)

1. Google Sheets Setup:
   • Create a new Google Sheet to store bug reports.
   • Define headers for the sheet (e.g., "Type," "Description by user").
2. Google Sheets Tool Configuration:
   • Read Tool: Configure the tool to read rows from the Google Sheet.
     • Connect a Google Sheets credential.
     • Select the document by ID (copy the ID from the URL).
     • Select the sheet (e.g., "Sheet1").
   • Append Tool: Configure the tool to append new rows to the Google Sheet.
     • Connect a Google Sheets credential.
     • Select the document by ID.
     • Select the sheet.
     • Use AI-generated mapping for each column mode.
3. Gmail Tool Configuration:
   • Configure the tool to send email notifications.
     • Connect a Gmail credential.
     • Set the "To" email address (the email that will receive the bug reports).
     • Use AI-generated subject and message content.

Testing and Deployment

1. Workflow Testing: Test the entire workflow or individual nodes to ensure proper functionality.
   • Use the test form URL to submit bug reports.
   • Pin data from test runs to work with it multiple times.
2. Error Handling: Address any errors that arise during testing.
3. Workflow Activation: Activate the AI agent to enable it to run.
4. Deployment: Deploy the NA10 workflow on a cloud server (Hostinger VPS is recommended) for autonomous operation.
   • Hostinger offers a one-click install template for NA10.
   • The KVM2 plan is recommended for running multiple AI agents.
   • Use coupon code "David" for a 10% discount.
5. Production URL: Use the production URL to collect feedback and run the AI agent in parallel.

Example Scenario and Results

1. Test Case: A user reports a UI bug: "There is no white mode."
2. Workflow Execution:
   • The bug report is submitted via the form.
   • The switch node routes the report to the UI bug agent.
   • The AI agent checks the Google Sheet for existing reports.
   • Since "There is no white mode" is a new bug, it's appended to the sheet.
   • An email notification is sent to the specified email address.
3. Verification:
   • The new bug report is added to the Google Sheet.
   • A new email is received with the bug report details.

Hostinger VPS for NA10 Deployment

• Recommendation: Hostinger VPS (KVM2 plan) is recommended for deploying NA10 AI agents due to its affordability and ease of use.
• Benefits:
  • One-click install template for NA10.
  • No limitations on the number of AI agents or executions.
  • Sufficient resources (8 GB RAM, 2 vCPU cores, 100 GB NVMe disk space, 8 TB bandwidth) for running hundreds of AI agents.
• Cost Savings: Hostinger VPS is more cost-effective than NA10 Cloud or other alternatives that charge per execution.

Key Arguments and Perspectives

• AI is Essential for Building AI Agents: Using AI (ChatGPT) to build AI agents significantly speeds up the development process compared to manual methods.
• Future-Proof Skill: Building AI agents is a valuable and future-proof skill due to the increasing automation of tasks and the potential for AI to replace entire departments.
• Customization is Key: AI agents should be customized to meet specific needs and use cases, rather than relying on generic examples.
• Importance of Reasoning Models: Using advanced reasoning models like GPT-3 (or GPT-4) is crucial for building complex and reliable AI agents.
• Deployment is Critical: An AI agent is useless if it's not deployed and running.

Notable Quotes

• "If you aren't using AI in order to build your AI agents, you're honestly falling behind." - David Andre
• "You don't need to be an NA10 expert, you don't even need to understand what JSON is, and you certainly don't need to be a programmer." - David Andre
• "We are living in the future. There's going to be like AI agents so advanced and so competent that they're not going to replace one person or like one position in the company, they're going to replace entire departments." - David Andre

Technical Terms and Concepts

• NA10: A workflow automation platform.
• AI Agent: A software program that uses artificial intelligence to perform tasks autonomously.
• JSON: A lightweight data-interchange format.
• System Prompt: A set of instructions given to an AI model to guide its behavior.
• OpenAI API Key: A secret key used to authenticate requests to the OpenAI API.
• GPT-3/GPT-4: Generative Pre-trained Transformer, a type of large language model developed by OpenAI.
• VPS: Virtual Private Server, a virtualized server that provides dedicated resources.
• KVM: Kernel-based Virtual Machine, a virtualization technology.
• NVMe: Non-Volatile Memory Express, a high-performance storage interface.
• LLM: Large Language Model

Logical Connections

The video demonstrates a step-by-step process of using ChatGPT to generate an NA10 workflow for bug report collection and management. It starts with setting up a project in ChatGPT, defining a system prompt, and providing example files. Then, it guides the viewer through the process of describing the desired bug reporting system to ChatGPT, generating the JSON workflow file, importing it into NA10, configuring the workflow nodes, setting up the AI agents, and integrating with Google Sheets and Gmail. Finally, it covers testing, deployment, and provides a real-world example to illustrate the functionality of the AI agent. The video emphasizes the importance of using AI to build AI agents, customizing the agents to specific needs, and deploying them on a cloud server for autonomous operation.

Data, Research Findings, or Statistics

• The video mentions that the KVM2 plan on Hostinger is sufficient to run hundreds or even thousands of NA10 agents.
• It also highlights the cost savings of using Hostinger VPS compared to NA10 Cloud or other alternatives that charge per execution.

Synthesis/Conclusion

The video provides a comprehensive guide on how to leverage ChatGPT to create NA10 workflows for automating tasks, specifically focusing on bug report collection and management. By using AI to generate the workflow, configuring the nodes, integrating with external services like Google Sheets and Gmail, and deploying the agent on a cloud server, users can significantly streamline their processes and improve efficiency. The video emphasizes the importance of AI in building AI agents and highlights the potential for AI to transform various aspects of business and management. The key takeaways are the ease of use, cost-effectiveness, and customization options offered by this approach.