Keyword Topic Extractor from Google Search

What This Agent Does

This Google Search Intelligence Agent automatically searches the web for topics you specify, extracts valuable keywords and insights from the results, and organizes everything into a spreadsheet—all without manual intervention. It's a powerful research automation tool that transforms raw search data into structured, actionable intelligence.

By combining Google Search with AI-powered analysis, this workflow saves you hours of manual research time. Instead of clicking through search results and manually cataloging information, the agent handles the entire process automatically. Each search result is analyzed by OpenAI's GPT-4o-mini model to extract relevant keywords and topics, then intelligently stored in your spreadsheet while automatically preventing duplicate entries.

Key benefits include:

• Automated research compilation – Search, analyze, and organize in minutes instead of hours
• AI-powered insights – Extract meaningful keywords and topics automatically using advanced language models
• Duplicate prevention – Smart logic ensures you never store the same URL twice
• Scalable intelligence gathering – Process hundreds of search results in a single workflow run
• Structured data output – All findings organized in a clean, searchable spreadsheet format

Ideal use cases:

• Competitive intelligence and market research
• SEO keyword research and topic clustering
• Content research for blog posts or articles
• Lead generation and prospect research
• Industry trend monitoring and analysis
• Academic research compilation

Who Is It For

This agent is perfect for marketing professionals, content creators, researchers, business analysts, and entrepreneurs who need to gather and organize web-based information quickly. Whether you're conducting competitive analysis, building content strategies, or monitoring industry trends, this workflow eliminates the tedious manual work of searching and categorizing results.

You don't need to be a technical expert—the workflow is designed to be intuitive and requires only basic configuration of your search parameters and spreadsheet destination.

Required Integrations

Google Search

Why it's needed: Google Search is the engine that powers your research. It retrieves relevant web results based on your search queries, providing the raw data that feeds into your analysis pipeline.

Setup steps:

1. Navigate to the Integrations section in your TaskAGI dashboard
2. Search for and select Google Search
3. Click Connect or Add Integration
4. You'll be directed to configure your Google Custom Search API credentials
5. Verify the connection by clicking Test Connection
6. Once confirmed, the integration is ready to use in your workflows

How to obtain API keys:

• Visit Google Cloud Console
• Create a new project or select an existing one
• Enable the Custom Search API
• Navigate to Credentials and create an API Key
• Also set up a Custom Search Engine at cse.google.com to get your Search Engine ID
• Copy both your API Key and Search Engine ID

Configuration in TaskAGI:

• Paste your API Key in the designated field
• Enter your Custom Search Engine ID
• Set your preferred number of results per search (typically 10-50)
• Save and test the connection

OpenAI

Why it's needed: OpenAI's GPT-4o-mini model analyzes each search result to intelligently extract keywords, topics, and key insights. This transforms raw search data into structured, meaningful information.

Setup steps:

1. Go to Integrations in TaskAGI and search for OpenAI
2. Click Connect to begin setup
3. You'll need your OpenAI API key (see below for obtaining it)
4. Paste your API key in the provided field
5. Select GPT-4o-mini as your model (pre-configured in this workflow)
6. Test the connection to ensure proper authentication

How to obtain API keys:

• Visit OpenAI Platform
• Sign up or log in to your account
• Navigate to API Keys in your account settings
• Click Create new secret key
• Copy the key immediately (it won't be shown again)
• Store it securely in your TaskAGI integration settings

Configuration in TaskAGI:

• Paste your API Key securely
• Confirm the model is set to gpt-4o-mini (optimal for cost and performance)
• Set your preferred temperature (0.7 is recommended for balanced creativity and consistency)
• Verify rate limits match your usage needs

Spreadsheet

Why it's needed: Your spreadsheet serves as the central repository for all extracted data. It stores URLs, keywords, topics, and metadata in an organized, queryable format that you can analyze, filter, and share.

Setup steps:

1. Open Integrations and select Spreadsheet
2. Choose your spreadsheet provider (Google Sheets, Excel Online, etc.)
3. Click Authorize and grant TaskAGI permission to access your spreadsheets
4. Once authorized, select the specific spreadsheet where results will be stored
5. Identify or create the target sheet/tab within that spreadsheet
6. Verify column headers are properly configured (see Configuration Steps below)

How to obtain credentials:

• For Google Sheets: Use your Google account; TaskAGI will request permission during authorization
• For Excel Online: Use your Microsoft account credentials
• Ensure the spreadsheet is accessible and you have edit permissions

Configuration in TaskAGI:

• Select your spreadsheet from the authorized list
• Choose the sheet name (e.g., "Search Results")
• Map columns for: URL, Keywords, Topics, Search Query, Date Added
• Enable header row recognition if your spreadsheet has headers
• Test write permissions by running a test execution

Configuration Steps

Step 1: Review Agent Information

The workflow begins with three information nodes that document the agent's purpose. These are for reference only—no configuration needed. They display:

• Agent Info: Overview of the automation
• Category Info: Classification and use case
• How It Works: Process explanation

Step 2: Set Default Values

This node initializes your workflow with essential parameters.

Configure:

• Search Query: Enter your primary search term (e.g., "AI automation tools")
• Max Results: Set how many search results to process (recommend 10-20 for initial tests)
• Spreadsheet Name: Specify your target spreadsheet name
• Sheet Name: Specify the target sheet within that spreadsheet

Example configuration:

Search Query: "sustainable packaging solutions"
Max Results: 15
Spreadsheet Name: "Market Research"
Sheet Name: "Competitor Analysis"

Step 3: Execute Google Search

This node retrieves search results based on your query.

Parameters:

• Query: Uses the value from Step 2
• Number of Results: Limits API calls and processing time
• Language: Set to your preferred language (default: English)

Data output: Returns a list of URLs, titles, and snippets from Google Search results.

Step 4: Loop Through Results

This node iterates through each search result individually.

How it works:

• Takes the search results array from Step 3
• Processes each result one at a time
• Passes individual result data to the next node
• Continues until all results are processed

No configuration needed—this node automatically handles iteration.

Step 5: Extract Keywords and Topics

This node uses OpenAI to analyze each search result.

Configure the prompt:

Extract keywords and topics from the following search result:
Title: [result_title]
Snippet: [result_snippet]
URL: [result_url]

Provide:
1. 5-7 relevant keywords (comma-separated)
2. 3-4 main topics (comma-separated)
3. Brief relevance assessment (1 sentence)

Parameters:

• Model: GPT-4o-mini (pre-configured)
• Temperature: 0.7 (balanced analysis)
• Max Tokens: 200 (sufficient for keyword extraction)

Output: Structured keywords and topics for each result.

Step 6: Check URL Exists

This node queries your spreadsheet to prevent duplicates.

Configure:

• Spreadsheet: Select your target spreadsheet
• Sheet: Select your target sheet
• Search Column: Set to "URL"
• Search Value: Uses the current result's URL

Purpose: Determines if this URL is already in your spreadsheet.

Step 7: Conditional Logic (If Not Exists)

This node routes the workflow based on the duplicate check.

Configure:

• Condition: "URL does not exist in spreadsheet"
• True Path: Proceed to insert the row (Step 8)
• False Path: Skip to no-operation node (Step 9)

Logic: Only new URLs are added; duplicates are skipped.

Step 8: Insert Row

When a URL is new, this node adds it to your spreadsheet.

Configure columns:

• URL: From current search result
• Title: From search result title
• Keywords: From OpenAI extraction
• Topics: From OpenAI extraction
• Search Query: From initial parameter
• Date Added: Current timestamp
• Source: "Google Search"

Example row:

URL: https://example.com/ai-tools
Title: Top AI Automation Tools 2024
Keywords: AI, automation, productivity, tools, enterprise
Topics: Artificial Intelligence, Business Automation, Software
Search Query: AI automation tools
Date Added: 2024-01-15
Source: Google Search

Step 9: Skip Duplicate

When a URL already exists, this no-operation node logs the skip.

Purpose: Maintains clean logs and prevents duplicate processing.

Testing Your Agent

Test Execution Steps

1. Prepare your environment:

• Ensure all three integrations are connected and tested
• Create or identify your target spreadsheet
• Verify spreadsheet columns match your configuration

2. Run a test execution:

• Click Test Run in the workflow editor
• Enter a simple test search query (e.g., "automation tools")
• Set Max Results to 5 for quick testing
• Click Execute

3. Monitor execution progress:

• Watch the workflow progress through each node
• Check for any error messages or warnings
• Verify that Google Search returns results
• Confirm OpenAI successfully extracts keywords

What to Verify at Each Step

Success Indicators

✅ Workflow completed successfully when:

• All nodes execute without errors
• Google Search returns relevant results
• OpenAI extracts meaningful keywords and topics
• New rows appear in your spreadsheet
• Duplicate URLs are properly skipped
• Execution time is reasonable (typically 30-60 seconds for 10 results)

✅ Verify spreadsheet contains:

• URLs from search results
• Extracted keywords (5-7 per result)
• Identified topics (3-4 per result)
• Original search query
• Timestamp of when data was added

Congratulations! Your Google Search Intelligence Agent is now fully operational and ready to automate your research workflow.