Part 1: Setting Up the Foundation in Make.com

First, you'll need a free account at Make.com. Once you're in, click "Create a new scenario" to open your canvas. Before we add our main tools, we need to set up a few variables to control our automation.

In your scenario, click the large purple "+" button, search for "Tools," and select the "Set Variable" module.

We'll create three variables to control our workflow.

1. The Time Filter (Start)

• Variable name: week_start

• Variable value: This JavaScript code will tell our system to only look for content published in the last 24 hours. javascript {{formatDate(addDays(now; -1); "YYYY-MM-DDTHH:mm:ss")}}

2. The Time Filter (End)

• Variable name: week_end

• Variable value: This sets the end of our 24-hour window. javascript {{formatDate(addDays(now; 0); "YYYY-MM-DDTHH:mm:ss")}}

3. The Content Sources

• Variable name: feed_urls

• Variable value: This is where we'll store our list of content sources using a format called JSON.

How to Find and Format Your Content Sources

To make our AI smart, we need to feed it high-quality information. A great way to do this is by using RSS feeds from top blogs and news sites in your industry.

1. Find Your Feeds: Go to Google and search for "[Your Industry] RSS feeds" + "feedspot". This will give you curated lists of top sources.

2. Collect the URLs: Copy the names and RSS feed URLs for 10-20 websites relevant to your audience.

3. Convert to JSON: Paste your list into ChatGPT with the following prompt to format it correctly.

   Convert the following into a JSON array where each object has the format: {name: string, url: string}.
   
   Input data:
   1. Name: Yes2next RSS Feed, 
   RSS Feed: https://yes2next.com/fitness-joy?format=rss 
   
   2. Name: Muscle & Fitness » Over 50 RSS Feed 
   RSS Feed: https://www.muscleandfitness.com/tag/over-50/feed/

4. Paste into Make.com: Copy the JSON output from ChatGPT and paste it into the feed_urls variable in your Make.com scenario.

What is JSON? Think of it as a universal address book. It organizes information into simple name: value pairs that computers can read instantly.

Part 2: Processing and Filtering Content

Now that our sources are ready, let's build the engine that processes them.

1. Parse the JSON

Add a "Parse JSON" module. This tool takes our single list of feed_urls and splits it into individual sources so we can check them one by one.

2. Retrieve New Articles with RSS

Add an "RSS" module and select "Retrieve RSS feed items."

• URL: Select the url variable from the JSON parser.

• Date from: Select your week_start variable.

• Date to: Select your week_end variable.

• Maximum items: Set to 10 to avoid overloading the system.

• Error Handling: Right-click the module and add an "Ignore" error handler. This makes our system more reliable by preventing it from stopping if one RSS feed is down.

What is RSS? It stands for Really Simple Syndication. It’s a standard way for websites to publish their latest content, like a newsletter subscription for your automation.

Part 3: Using AI to Select the Best Content

This is where the magic happens. We'll use an AI model to act as our content curator, analyzing each article and deciding if it's a good fit for our audience.

1. Add the "Selector AI" (OpenRouter)

Add an "OpenRouter" module. You'll need a free account and an API key from OpenRouter.ai.

• Model: Choose a fast model like GPT-4o-mini.

• Role: User

• Message: Paste the "Selection Prompt" below. This prompt acts as the AI's brain, telling it exactly how to evaluate each article.

  Analyze this article and determine if it's valuable for small and medium businesses:
  
  Title: {{15.rssFields.title}}
  Description: {{15.rssFields.description}}
  
  Evaluate based on these criteria:
  1. Business Impact: Does this help SMBs improve operations, save costs, or increase revenue?
  2. Accessibility: Is this technology/concept accessible to businesses without large tech teams?
  3. Actionability: Can SMBs actually implement or benefit from this information? (Short or near long term)
  4. Relevance: Is this relevant to common SMB challenges (productivity, customer service, marketing, etc.)?
  
  Respond ONLY with valid JSON in this exact format:
  
  {
    "decision": "SELECTED" or "REJECTED",
    "score": 1-10,
    "reason": "Brief explanation (1-2 sentences)",
    "category": "productivity" or "marketing" or "customer_service" or "operations" or "finance" or "technology" or "other",
    "business_impact": "high" or "medium" or "low",
    "implementation_difficulty": "easy" or "medium" or "hard"
  }
  
  Do not include any text outside the JSON response
  
  

2. Parse the AI's Decision

Add another "Parse JSON" module. This will read the AI's decision (SELECTED or REJECTED) so we can use it in the next step.

Part 4: Routing, Summarizing, and Scripting

Our system now knows which articles are gold and which are junk. Let's separate them and turn the good ones into content.

1. Add a Router

A "Router" module splits the workflow into different paths. We'll create two paths: one for rejected content and one for selected content.

• Route 1 (Rejected): Set the filter to decision equals REJECTED. Connect this to a "Google Sheets" module to log the rejected articles for later analysis.

• Route 2 (Selected): Set the filter to decision equals SELECTED. This path continues to the next AI modules.

2. Create the AI Summarizer

Add another "OpenRouter" module to the "Selected" path. This AI's job is to create a detailed business summary of the article.

• Model: GPT-4o-mini

• Prompt: Use the detailed "Summarization Prompt" you've prepared. This prompt asks the AI to extract key benefits, action items, and a suggested title.

3. Create the AI Script Writer

Add a final "OpenRouter" module. This is our creative genius.

• Model: Use a creative model like Claude 3.5 Sonnet or Gemini 1.5 Pro.

• Prompt: Use your "Script Writing Prompt" that includes the 6-Element Dopamine Script Structure. This will turn the business summary into a psychologically engaging, 45-60 second video script.

4. Parse and Save the Final Output

Add one last "Parse JSON" module to read the scriptwriter's output. Then, connect it to a "Google Sheets" module ("Selected" tab) and map the fields:

• Title -> Title

• Summary -> Summary

• suggestShortsTitle -> Script Title

• script -> Script

Part 5: Run Your Automation and Customize

Your complete automation is built. Hit the "Run once" button and watch as your Google Sheet populates with perfectly curated content and ready-to-film viral scripts.

Customization and Next Steps

• Adjust the Time Window: Want to scan content from the last 3 days? Simply change the -1 in your week_start variable to -3.

• Fine-Tune the AI: If the AI is rejecting good articles, analyze the "Rejected" tab in your Google Sheet and adjust the "Selection Prompt" to be more specific.

• Scale Your Sources: Don't be afraid to add 50 or even 100 RSS feeds to the feed_urls variable. The more high-quality data you provide, the better your results will be.

You now have a fully automated content research and creation system that works for you 24/7, freeing you up to focus on what matters most: growing your business.