Software Vulnerability Patent Tracking AI Agent

What This Agent Does

This intelligent automation agent monitors Google Patents for new patent filings relevant to your business, automatically filtering and alerting you to the most important discoveries. Instead of manually searching patent databases weekly, this workflow runs on autopilot—fetching patent data, using AI to extract structured information, analyzing relevance to your criteria, and notifying your team via Slack while storing valuable leads in Intercom for follow-up.

Key benefits include:

• Save 5+ hours weekly on manual patent research and monitoring
• Never miss critical patents with automated weekly checks
• AI-powered filtering ensures you only see patents that matter to your business
• Instant team notifications via Slack for time-sensitive opportunities
• Automatic CRM integration stores patent contacts in Intercom for relationship building

This workflow is perfect for competitive intelligence, prior art research, partnership discovery, technology scouting, and IP portfolio management.

Who Is It For

This automation is ideal for:

• IP and Legal Teams who need to monitor competitor patent filings and identify potential infringement risks
• R&D Departments tracking emerging technologies and innovation trends in their field
• Business Development Professionals seeking partnership opportunities and licensing prospects
• Product Managers conducting competitive analysis and market research
• Innovation Scouts at corporations identifying acquisition targets or technology trends
• Patent Attorneys performing prior art searches and freedom-to-operate analyses

Whether you're a startup monitoring your competitive landscape or an enterprise managing a large IP portfolio, this agent delivers actionable patent intelligence directly to your team.

Required Integrations

Anthropic (Claude AI)

Why it's needed: Claude AI powers the intelligent extraction of patent data from HTML pages, transforming unstructured web content into clean, structured patent information including titles, inventors, filing dates, and abstracts.

Setup steps:

1. Visit console.anthropic.com and create an account or sign in
2. Navigate to API Keys in your account settings
3. Click Create Key and provide a descriptive name like "TaskAGI Patent Monitor"
4. Copy the API key immediately (it won't be shown again)
5. In TaskAGI, go to Integrations → Anthropic
6. Click Add Connection and paste your API key
7. Test the connection to verify authentication

Important: Anthropic charges per token used. This workflow uses the Claude Sonnet 4 model, which offers excellent accuracy at moderate cost. Expect approximately $0.50-2.00 per weekly run depending on patent result volume.

Slack

Why it's needed: Slack delivers real-time notifications when relevant patents are discovered, ensuring your team can act quickly on time-sensitive opportunities or competitive threats.

Setup steps:

1. In your Slack workspace, go to Settings & Administration → Manage Apps
2. Search for "TaskAGI" or create a custom app at api.slack.com/apps
3. Click Create New App → From Scratch
4. Name it "Patent Monitor" and select your workspace
5. Navigate to OAuth & Permissions in the sidebar
6. Under Scopes, add the chat:write bot token scope
7. Click Install to Workspace and authorize the app
8. Copy the Bot User OAuth Token (starts with xoxb-)
9. In TaskAGI, go to Integrations → Slack
10. Click Add Connection and paste your OAuth token
11. Select the default channel where alerts should be posted

Pro tip: Create a dedicated #patent-alerts channel to keep notifications organized and ensure the right team members are subscribed.

Intercom

Why it's needed: Intercom stores patent inventor and assignee information as contacts, enabling your team to track relationships, add notes, and manage outreach for licensing, partnerships, or competitive intelligence.

Setup steps:

1. Log into your Intercom workspace at app.intercom.com
2. Click the settings gear icon → Developers → Developer Hub
3. Navigate to Authentication → Access Tokens
4. Click New access token and name it "TaskAGI Patent Integration"
5. Grant Read and Write permissions for Contacts
6. Copy the generated access token
7. In TaskAGI, go to Integrations → Intercom
8. Click Add Connection and paste your access token
9. Enter your Intercom workspace ID (found in your workspace URL)
10. Test the connection to verify proper authentication

Note: Contacts created by this workflow will appear in your Intercom dashboard with custom attributes for patent details, making them easy to segment and manage.

Configuration Steps

1. Weekly Patent Check (Trigger)

Configure when your patent monitoring runs:

• Interval Type: Select Weekly
• Day of Week: Choose Monday (or your preferred day)
• Time: Set to 09:00 for morning delivery
• Timezone: Select your team's timezone

This ensures fresh patent alerts arrive at the start of your work week when your team is ready to act on them.

2. Construct Patent Query (Function)

Build your Google Patents search query:

// Example configuration
const searchTerms = "machine learning medical diagnosis";
const dateRange = "after:2024-01-01";
const query = `${searchTerms} ${dateRange}`;
return { query, url: `https://patents.google.com/?q=${encodeURIComponent(query)}` };

Customize these parameters:

• searchTerms: Your technology keywords (e.g., "quantum computing", "CRISPR gene editing")
• dateRange: Filter by filing date (e.g., "after:2024-01-01" for recent patents)
• assignee: Add company filters like "assignee:Microsoft"
• classification: Use CPC codes for precise targeting (e.g., "cpc:H04L")

3. Fetch Patents Page (HTTP Request)

Retrieves the Google Patents search results page:

• Method: GET
• URL: Use the URL from the previous node: {{4132.url}}
• Headers: Add User-Agent: Mozilla/5.0 to avoid blocking
• Timeout: Set to 30000 ms (30 seconds)

The response body contains the HTML that will be analyzed by Claude AI.

4. Extract Patent Data (Claude AI)

Configure Claude to parse patent information:

• Model: claude-sonnet-4-5-20250929 (pre-selected for optimal performance)
• Prompt:

Extract all patent listings from this Google Patents search results page. For each patent, provide:
- Patent number
- Title
- Inventors (comma-separated)
- Assignee/Company
- Filing date
- Abstract (first 200 characters)

Return as a JSON array of objects.

• Input: {{4133.body}} (the HTML from the previous step)
• Max Tokens: 4000 (sufficient for ~10-15 patents)

Claude will return structured JSON data that subsequent nodes can process.

5. Analyze Relevance (Function)

Score each patent's relevance to your business criteria:

// Example relevance scoring logic
function scorePatent(patent) {
  let score = 0;
  
  // High-value assignees
  const competitors = ["Google", "Microsoft", "Amazon"];
  if (competitors.some(c => patent.assignee.includes(c))) score += 30;
  
  // Key technology indicators
  const keywords = ["neural network", "deep learning", "AI"];
  keywords.forEach(kw => {
    if (patent.title.toLowerCase().includes(kw)) score += 20;
  });
  
  // Recent filings are more valuable
  const filingDate = new Date(patent.filingDate);
  const monthsOld = (Date.now() - filingDate) / (1000 * 60 * 60 * 24 * 30);
  if (monthsOld < 3) score += 15;
  
  return score;
}

return patents.map(p => ({ ...p, relevanceScore: scorePatent(p) }));

Customize scoring criteria based on your priorities: competitor activity, specific technologies, inventor reputation, or patent classification.

6. Relevant? (Conditional Branch)

Filter patents based on relevance score:

• Condition: {{4135.relevanceScore}} greater than 50
• True Path: Proceeds to Slack alert and Intercom storage
• False Path: Skips notification (patent logged but not alerted)

Adjust the threshold (50) based on your alert volume preferences. Higher thresholds = fewer, more targeted alerts.

7. Slack Alert (Notification)

Send formatted alerts to your team:

• Channel: Select your #patent-alerts channel
• Message Template:

🔔 *New Relevant Patent Detected*

*{{4135.title}}*
Patent #: {{4135.patentNumber}}
Assignee: {{4135.assignee}}
Inventors: {{4135.inventors}}
Filing Date: {{4135.filingDate}}
Relevance Score: {{4135.relevanceScore}}/100

Abstract: {{4135.abstract}}

<https://patents.google.com/patent/{{4135.patentNumber}}|View Full Patent>

Formatting tips: Use Slack markdown for emphasis, include direct links, and add emoji for visual scanning.

8. Map Fields for Storage (Data Transformation)

Prepare patent data for Intercom's contact format:

• email: {{4135.inventors[0]}}.patent@placeholder.com (Intercom requires email)
• name: {{4135.inventors[0]}}
• custom_attributes:
  • patent_number: {{4135.patentNumber}}
  • patent_title: {{4135.title}}
  • assignee: {{4135.assignee}}
  • filing_date: {{4135.filingDate}}
  • relevance_score: {{4135.relevanceScore}}

Note: Since inventors rarely have public emails, we use placeholder addresses. The real value is in the custom attributes for tracking and segmentation.

9. Save to Intercom (CRM Storage)

Create or update contact records:

• Operation: Create Contact
• Data Source: {{4138}} (mapped fields from previous node)
• Duplicate Handling: Update if exists (based on email)

Contacts appear in Intercom with all patent metadata, enabling your team to add notes, set reminders, and manage follow-up activities.

Testing Your Agent

Initial Test Run

1. Click Test Workflow in the top-right corner of the TaskAGI editor
2. The trigger will be simulated, starting the workflow immediately
3. Watch each node execute in real-time—green checkmarks indicate success

Verification Checklist

After Construct Patent Query:

• Verify the generated URL is valid
• Check that search parameters are correctly encoded
• Confirm date ranges and keywords are as expected

After Fetch Patents Page:

• Inspect the response body to ensure HTML was retrieved
• Status code should be 200
• Body should contain patent listing HTML (not an error page)

After Extract Patent Data:

• Review the JSON output for proper structure
• Verify all expected fields are populated (title, inventors, etc.)
• Check that 5-15 patents were extracted (typical result count)

After Analyze Relevance:

• Examine relevance scores to ensure they align with your criteria
• Adjust scoring logic if scores seem too high/low
• Confirm at least one patent exceeds your threshold

After Slack Alert:

• Check your Slack channel for the notification
• Verify formatting displays correctly
• Click the patent link to ensure it's valid

After Save to Intercom:

• Open Intercom and search for the newly created contact
• Verify all custom attributes populated correctly
• Check that the contact appears in your expected segment

Expected Results

A successful test run should:

• Complete in 30-60 seconds total
• Extract 5-15 patents from Google Patents
• Score each patent with your relevance algorithm
• Send 1-3 Slack notifications (for patents above threshold)
• Create 1-3 Intercom contacts with complete metadata

Troubleshooting

"HTTP Request Failed" Error

Cause: Google Patents may block automated requests or the search returned no results.

Solutions:

• Add a User-Agent header mimicking a real browser
• Reduce request frequency if running tests repeatedly
• Verify your search query returns results when entered manually on Google Patents
• Add a 2-3 second delay before the HTTP request node

"Claude Extraction Returned Empty Array"

Cause: The HTML structure changed or Claude couldn't parse the content.

Solutions:

• Inspect the HTTP response body to verify it contains patent listings
• Update the Claude prompt with more specific HTML element instructions
• Try increasing max_tokens to allow more processing capacity
• Check that Google Patents didn't return a CAPTCHA page

"Relevance Scores All Zero"

Cause: Your scoring logic doesn't match the actual patent data structure.

Solutions:

• Log the patent object structure to see available fields
• Verify field names match exactly (case-sensitive)
• Test your scoring function with sample patent data
• Add default scoring for patents that don't match specific criteria

"Slack Message Not Appearing"

Cause: Insufficient permissions or incorrect channel configuration.

Solutions:

• Verify the TaskAGI bot is invited to your target channel
• Check that chat:write scope is enabled in Slack app settings
• Test with a public channel before using private channels
• Confirm the channel ID or name is correctly specified

"Intercom Contact Creation Failed"

Cause: Missing required fields or invalid data format.

Solutions:

• Ensure email field is present (even if placeholder)
• Verify custom attribute names match your Intercom setup
• Check that date fields are in ISO 8601 format
• Confirm your API token has write permissions for contacts

Next Steps

After Successful Setup

1. Monitor your first few runs to ensure alert quality matches expectations
2. Adjust the relevance threshold based on alert volume (raise to reduce noise, lower to catch more)
3. Create Intercom segments for different patent categories (competitor patents, technology areas, etc.)
4. Set up Slack channel notifications so team members are alerted to new messages
5. Document your scoring criteria so team members understand why patents are flagged

Optimization Suggestions

Refine Your Search Query:

• Add multiple search variations to catch different terminology
• Use CPC classification codes for precision targeting
• Exclude noise with negative keywords (e.g., "-toy -game")

Enhance Relevance Scoring:

• Incorporate inventor reputation scores from external databases
• Weight recent patents more heavily for competitive intelligence
• Add citation analysis to identify foundational patents

Expand Notifications:

• Send high-priority patents to a separate urgent-alerts channel
• Create weekly digest emails summarizing all patents found
• Add Microsoft Teams or email notifications for stakeholders not on Slack

Enrich Intercom Data:

• Add tags to contacts based on patent technology area
• Create automated sequences for outreach to inventors
• Link patents to company records for portfolio tracking

Advanced Usage Tips

Multi-Query Monitoring: Duplicate this workflow for different technology areas, each with customized search terms and relevance criteria.

Competitive Intelligence Dashboard: Export Intercom data to a BI tool for trend analysis and competitor patent velocity tracking.

Integration with IP Management Systems: Connect the workflow to PatSnap, Derwent, or internal IP databases for comprehensive portfolio management.

Automated Prior Art Searches: Modify the workflow to run on-demand when your team files a new patent application, automatically identifying potential prior art.

Partnership Pipeline: Use Intercom's CRM features to track outreach to patent assignees for licensing discussions or acquisition opportunities.

---

You're now ready to automate your patent monitoring! This workflow will save your team hours of manual research while ensuring you never miss important patent developments in your industry. As you gain experience, continue refining your relevance criteria to maximize signal and minimize noise.