Build AI Agents for anything & everything - here’s how

Key Concepts

• AI Agents: Autonomous programs designed to perform specific tasks.
• MCP (Model Composition Protocol): A protocol that allows different AI models and tools to communicate and work together.
• NA10: An open-source, visual platform for building and automating workflows, now integrated with MCP.
• Cloud Desktop: An application that can interact with MCP-enabled AI agents.
• Hostinger: A web hosting provider used to deploy and run AI agents 24/7.
• Workflows: Automated sequences of actions within NA10.
• API Keys: Codes used to authenticate and authorize access to services like OpenAI.
• Google Cloud Console: A platform for managing Google Cloud services, including API access and credentials.
• Oauth 2.0: An authorization framework that enables applications to obtain limited access to user accounts on an HTTP service, such as Google.

Building AI Agents with MCP and NA10

Introduction

The video focuses on how to build AI agents using NA10 and MCP, enabling automation of various tasks. It emphasizes the importance of deploying these agents on a server for continuous operation. The speaker, David Andre, highlights the integration of MCP into NA10 as a significant advancement, allowing for the creation of AI agents that can interact with a wide range of tools and services.

Why NA10?

NA10 is chosen for its:

1. Open-source nature: Can be run anywhere.
2. Visual interface: Easy to understand workflows.
3. Ease of setup and use.
4. Customizability: Powerful AI agent capabilities.

Combining MCP and NA10

The latest NA10 update introduces the MCP server trigger, enabling:

1. Running NA10 from anywhere: AI agents using MCP (e.g., Cloud Desktop, Cursor) can trigger NA10 workflows.
2. Bringing any MCP tool into NA10: Access to thousands of MCP tools within NA10 automations.

The Biggest Mistake: Failure to Deploy

The most common mistake is building AI agents but failing to deploy them on a server to run 24/7. Real value comes from autonomous agents that save time continuously.

Hosting AI Agents on Hostinger

Hostinger is recommended for hosting AI agents due to its ease of use and affordability. The KVM2 plan is sufficient for hosting hundreds of AI agents on NA10.

• Setup: Select a plan, choose a period (12 months recommended), and use the code "David" for an additional 10% discount.
• NA10 Setup: Select the server closest to you, choose the pre-built NA10 template, and set a root password.

Updating NA10 to the Latest Version

To access the MCP server trigger, NA10 must be updated. This involves using the browser terminal in Hostinger and running specific Docker commands:

1. clear (clears the terminal)
2. curl -fSL https://get.docker.com -o get-docker.sh
3. sh get-docker.sh (installs Docker)
4. systemctl enable docker (starts Docker on reboot)
5. docker compose pull n8n (pulls the latest NA10 image)
6. docker compose down (stops and removes the old container)
7. docker compose up -d (starts NA10 with the new Docker image)
8. docker ps (checks if everything is running)

Connecting NA10 to Cloud Desktop

1. Install Cloud Desktop: Download and install the application.

2. Enable Developer Mode: In Cloud Desktop settings (accessed via Mac OS settings), enable developer mode and edit the cloth_desktop_config.json file.

3. Update Config File: Replace the URL in the JSON file with the production URL from the NA10 MCP server trigger. The JSON structure is:

   {
     "n8n": {
       "url": "YOUR_N8N_PRODUCTION_URL"
     }
   }
   

4. Install NodeJS: If the config is not loading, install NodeJS from nodejs.org.

5. Restart Cloud Desktop: Completely quit and restart the application.

Building a Simple Calculator Tool

1. In NA10, add an MCP server trigger.
2. Switch to production URL and save.
3. Add a calculator tool (no setup required).
4. In Cloud Desktop, the calculator tool from NA10 should appear.

Creating a Google Sheets Integration

This section details creating a more complex integration using Google Sheets.

1. Create a New Workflow: In NA10, create a new workflow triggered by "When executed by another workflow."
2. Add an AI Agent: Add an AI agent node to the workflow.
3. Configure OpenAI Connection:
   • Create a new credential using an OpenAI API key.
   • Obtain the API key from platform.openai.com.
4. Add a Google Sheets Tool:
   • Add a Google Sheets node to the workflow.
   • Create a new credential using OAuth 2.0.
   • Obtain the client ID and client secret from the Google Cloud Console.
5. Google Cloud Console Setup:
   • Create a new project in the Google Cloud Console.
   • Enable the Google Sheets API and Google Drive API.
   • Configure the OAuth consent screen.
   • Create OAuth 2.0 credentials (client ID and client secret).
   • Add the NA10 redirect URL to the authorized redirect URIs.
6. Connect Google Account: In NA10, sign in with your Google account and grant the necessary permissions.
7. Select Google Sheet: Choose the Google Sheet and sheet within the document.
8. Configure Append Row: Set the operation to "Append Row."
9. Set up MCP Server Trigger:
   • In the MCP server trigger, add a tool to call the NA10 workflow.
   • In the Google Sheets tool, provide a description without special characters.
   • In the instructions, define the description by AI.
10. Test the Integration: Use Cloud Desktop to send a prompt to add mock data to the Google Sheet.

Key Takeaways

• Specialized AI Agents: Create specialized AI agents for specific tasks and delegate to them from a main AI agent.
• Data Propagation: Ensure data flows correctly from the MCP server trigger through the tools and into the secondary workflow.
• Error Handling: Use AI tools like ChatGPT and 03 to troubleshoot errors and find solutions.
• Continuous Deployment: Deploy AI agents on a server like Hostinger to run 24/7 and automate tasks continuously.

Conclusion

The video provides a detailed guide on building and deploying AI agents using NA10, MCP, and Hostinger. It emphasizes the importance of continuous deployment and highlights the potential for automating various tasks, such as lead generation and data entry. By following the steps outlined in the video, viewers can create AI agents that save time and improve productivity.