6 simple AI Agents you must have - beginners guide

Key Concepts

• AI Agents: Automated workflows built using platforms like NA10 to perform specific tasks.
• NA10: A low-code/no-code platform for building and hosting AI agents.
• Triggers: Events that initiate an AI agent workflow (e.g., scheduled time, email received).
• Nodes: Individual steps or actions within an NA10 workflow (e.g., Gmail integration, AI agent, code execution).
• LLMs (Large Language Models): AI models like GPT-4 used within AI agents for natural language processing and decision-making.
• System Prompt: Instructions given to an LLM to define its behavior, task, and output format.
• User Prompt: Input provided to an LLM, containing specific data or questions for it to process.
• Replicate: A platform for hosting and accessing AI models, particularly for video and image generation.
• Hostinger: A web hosting provider used to host NA10 instances and run AI agents.
• VPS (Virtual Private Server): A virtualized server used for hosting applications like NA10.

1. Infinite Lead Follow-Up Agent

• Main Topic: Automating email follow-ups to leads until a response or sale is achieved.
• Trigger: Scheduled trigger (e.g., once a day).
• Steps:
  1. Gmail Get Many Messages: Retrieves sent emails from a specified Gmail inbox using the "in:sent" filter.
  2. Code Node (JavaScript): Processes the retrieved emails to identify those that haven't received a response. Uses ChatGPT to generate the code.
  3. Loop Over Items: Iterates through the list of emails without responses.
  4. AI Agent: Generates a personalized follow-up message based on the email subject and snippet. Uses a system prompt to define the agent's behavior and tone.
  5. Gmail Reply to a Message: Sends the generated follow-up email.
• Technical Details:
  • Setting up Gmail API in Google Cloud Console: Creating a project, enabling the Gmail API, configuring the OAuth consent screen, and creating client credentials (client ID and client secret).
  • Using the "in:sent" filter in the Gmail node to retrieve sent emails.
  • Using JavaScript code to filter emails based on response status.
  • Using JSON expressions to pass data between nodes.
• Key Argument: This agent saves time by automating a repetitive task, ensuring no leads are forgotten.
• Quote: "This agent is universal. Everyone needs him."

2. Realistic AI Videos Agent

• Main Topic: Generating realistic AI videos from text prompts using the V3 model on Replicate.
• Trigger: Scheduled trigger or other event (e.g., WhatsApp, Telegram).
• Steps:
  1. Google Sheets Get Rows: Retrieves video ideas and prompts from a Google Sheet.
  2. AI Agent: Refines the video idea into an optimized prompt for the V3 model.
  3. HTTP Request: Sends a request to the Replicate API with the prompt and other parameters.
  4. HTTP Request (Streaming): Streams the video output from Replicate.
  5. Google Drive Upload File: Uploads the generated video to Google Drive.
  6. Google Sheets Update Row: Updates the Google Sheet with the video link and status.
• Technical Details:
  • Using the Replicate API: Obtaining an API key, constructing the HTTP request with headers (authorization, content type), and specifying the request body in JSON format.
  • Using the V3 model for realistic video generation.
  • Handling new line characters in the prompt to avoid errors with Replicate.
• Real-World Applications: Marketing videos, B-roll footage, promotions, interviews, upskilling employees, ASMR videos.
• Quote: "Realistic videos are changing the world."

3. YouTube Channel Performance Tracker Agent

• Main Topic: Tracking a YouTube channel's performance by monitoring video views over time.
• Trigger: Scheduled trigger (e.g., every 11 hours).
• Steps:
  1. RSS Feed: Retrieves the RSS feed of a YouTube channel.
  2. Code Block (JavaScript): Filters out videos published more than 12 hours ago.
  3. Google Sheets Append Row: Adds new videos to a Google Sheet with a "start" status.
  4. Google Sheets Get Rows: Retrieves rows from the Google Sheet with a "start" status.
  5. Edit Fields: Sets the row number and video link.
  6. YouTube Get Video: Retrieves video views based on the video ID.
  7. Google Sheets Update Row: Updates the Google Sheet with the video views for day zero.
  8. Wait: Pauses the workflow for one day.
  9. Repeat Steps 6-8: Tracks video views and updates the Google Sheet for days 1, 2, 3, 7, 14, and 30.
• Technical Details:
  • Using the YouTube API to retrieve video views.
  • Using the "Wait" node to pause the workflow for a specified duration.
  • Tracking video performance over time to identify outliers.
• Real-World Applications: Monitoring competitor channels, tracking personal channel performance, identifying successful video ideas.
• Quote: "Lots of YouTubers would pay some serious money to have it."

4. Expense Tracker Agent

• Main Topic: Automatically tracking expenses from email receipts and invoices.
• Trigger: Gmail node triggered by new messages received.
• Steps:
  1. Gmail: Monitors a specified Gmail inbox for new messages.
  2. AI Agent: Analyzes the email content to determine if it's an expense-related email (invoice, receipt).
  3. Google Sheets Append Row: Appends a row to a Google Sheet with the date, transaction name, cost, category, and relevance score.
• Technical Details:
  • Using an AI agent to classify emails as expenses or not.
  • Categorizing expenses and assigning a relevance score.
• Real-World Applications: Automating accounting tasks for individuals and small businesses.
• Quote: "It automates most of your accounting needs."

5. Personalized Outreach Email Agent

• Main Topic: Generating personalized outreach emails for potential clients or contacts.
• Trigger: Scheduled trigger.
• Steps:
  1. Google Sheets Get Rows: Retrieves contact information from a Google Sheet with a "start" status.
  2. Perplexity or SER API: Performs web research on the contact person and company.
  3. AI Agent (Research): Researches on the person and the company to gather information for personalization.
  4. AI Agent (Email Writing): Writes a personalized outreach email based on the research results.
  5. Google Sheets Update Row: Updates the Google Sheet with the generated email and a "pending approval" status.
  6. Manual Approval: A human reviews and approves the email.
  7. Google Sheets Get Rows (Confirmed): Retrieves rows from the Google Sheet with a "confirmed" status.
  8. Gmail Send Message: Sends the approved email.
  9. Google Sheets Update Row (Sent): Updates the Google Sheet with a "sent" status.
• Technical Details:
  • Using Perplexity or SER API for web research.
  • Using two AI agents: one for research and one for email writing.
  • Implementing a manual approval step to ensure email quality.
• Real-World Applications: Cold outreach, sales, networking.
• Key Argument: Personalization improves email deliverability, open rates, and response rates.

6. Birthday Reminder Agent

• Main Topic: Reminding users of upcoming birthdays and generating gift ideas.
• Trigger: Scheduled trigger (e.g., once a week).
• Steps:
  1. Google Sheets Get Rows: Retrieves birthday information from a Google Sheet.
  2. AI Agent: Brainstorms gift ideas based on the person's name, gender, and interests.
  3. Tool Structured Output Parser: Formats the output of the agent.
  4. If/Else Node: Checks if the birthday is within 7 days.
  5. Telegram Send Message: Sends a text message on Telegram reminding the user of the upcoming birthday and providing gift ideas.
• Technical Details:
  • Using an AI agent to brainstorm gift ideas.
  • Using an if/else statement to conditionally send a reminder.
• Real-World Applications: Personal use, improving relationships.
• Key Argument: This agent helps users remember birthdays and find appropriate gifts, strengthening relationships.

Turning AI Agents into a Profitable Side Hustle

• Main Topic: Monetizing AI agent building skills by offering specialized services to businesses.
• Key Points:
  • Specialize: Focus on solving specific problems for a particular niche (e.g., dentists, real estate agents).
  • Irresistible Offer: Offer a free AI consultation to demonstrate expertise and build trust.
  • Profit Maximization: Add expensive but valuable add-ons (e.g., fast delivery, done-for-you code, code guidance).
  • Understand the Target Avatar: Know their problems, dream outcomes, and what they value.
• Examples of Add-ons:
  • Fast delivery (additional fee).
  • Done-for-you code (additional fee).
  • Code guidance (additional fee).
• Quote: "It's not about creating the most complicated agent. Far from it, guys. All you need is a reliable agent that specifically targets the business and helps to solve a specific problem for that business."

Hosting AI Agents

• Main Topic: Using Hostinger to host NA10 instances and run AI agents.
• Key Points:
  • Hostinger offers a one-click install template for NA10.
  • The KVM2 plan is recommended for running hundreds of AI agents.
  • Using the coupon code "David" provides a 10% discount.
  • Hostinger provides data on server usage, which is not available on NA10 Cloud.

Conclusion

The video provides a comprehensive guide to building six essential AI agents using NA10, covering various applications from lead follow-up to expense tracking and personalized outreach. It emphasizes the importance of understanding the underlying logic, using code where appropriate, and specializing in a niche to monetize AI agent building skills. The video also highlights Hostinger as a simple and affordable solution for hosting NA10 instances. The main takeaway is that AI agents can significantly improve productivity and efficiency in both personal and professional life, and that building and selling these agents can be a profitable side hustle.