🧠 Universal AI Semantic Monitor

Turn the web into a programmable boolean sensor.

This Actor does the work of a 24/7 research analyst. Instead of just scraping text, it reads websites using state-of-the-art AI to answer a specific Yes/No question based on your intent.

1. 🕵️ The Scout: Monitors specific URLs or auto-discovers relevant pages via DuckDuckGo.
2. 🧠 The Analyst: Uses LLMs (GPT-4o, Claude 3.5, Gemini 1.5, or Llama 3) to "fact-check" page content against your question.
3. 🔔 The Messenger: Alerts you immediately via Email or Webhook only when the answer is YES.

---

🎯 Perfect For

• 📉 Market Intelligence – "Has competitor X released feature Y yet?"
• 🎮 Gaming & Tech – "Is GTA 6 released?" or "Is the new iPhone available for pre-order?"
• 📰 Crisis Monitoring – "Is there a confirmed outage in Region Z?"
• ⚖️ Legal & Compliance – "Does this policy page mention GDPR compliance?"
• 🤖 Event Triggers – Fire a webhook to your API only when a specific real-world event is confirmed.

---

⚙️ Key Features

🧠 Universal AI Adapter (BYOK)

Bring Your Own Key (BYOK) for maximum flexibility. The Actor supports all major AI providers. You can choose the model that fits your budget and speed requirements:

• Groq: Ultra-fast inference (Llama 3.1, Mixtral)
• OpenAI: Industry standard (GPT-4o, GPT-4o-mini)
• Anthropic: High reasoning capability (Claude 3.5 Sonnet, Haiku)
• Google Gemini: Large context windows (Gemini 1.5 Flash/Pro)

🕵️ Auto-Discovery Mode

Don't have a list of URLs? No problem.

• If you leave the manualUrls field empty, the Actor automatically searches DuckDuckGo for your intent.
• It finds the top most relevant pages and monitors them for you.

✅ Boolean Fact-Checking Logic

Unlike standard scrapers that dump raw text, this Actor performs semantic analysis.

• It scrapes the visible text of the page.
• It prompts the AI: "Based ONLY on this text, is the answer to the user's question YES?"
• It outputs a structured eventDetected: true/false result.

🔔 Smart Alerts & Snapshots

• Change Detection: Uses MD5 hashing to ignore pages that haven't changed since the last run.
• Visual Proof: If a match is found (eventDetected: true), it captures a screenshot (sent via email).
• Webhooks: Pushes a JSON payload to your server immediately upon detection.

---

🎛️ Input Configuration

Section	Setting	Description
🎯 Core	Intent	The specific question to answer (e.g., "Is the store open?", "Is inflation down?")
🌐 Sources	Manual URLs	Specific list of URLs to monitor. If empty, auto-discovery is used.
⚡ AI Engine	API Keys	Provide at least one key: openaiApiKey, anthropicApiKey, groqApiKey, or geminiApiKey.
🧠 Models	Model Selection	Select specific models (e.g., gpt-4o-mini, llama-3.1-8b, claude-3-haiku).
🔔 Alerts	Send Email	Check this to receive an email with a screenshot when a match is found.
🔔 Alerts	Webhook URL	Optional URL to receive a POST request with the analysis result.

---

📦 Outputs

1) Dataset Results

The Actor pushes a simplified analysis record for every URL checked.

• Status: MATCH FOUND or NO MATCH.
• Summary: A one-sentence explanation generated by the AI.
• EventDetected: Boolean (true/false).
• Timestamp: Time of the check.

2) Webhook Payload

If a Webhook URL is provided and an event is detected, you receive:

{
  "eventDetected": true,
  "summary": "The article confirms that the product was released on January 15th.",
  "url": "https://example.com/news/product-launch",
  "status": "MATCH FOUND"
}

3) Email Notifications

If enabled, you will receive an email from Apify with:

• Subject: 🔔 True: [Summary of the finding]
• Body: Link to the URL and the AI's reasoning.
• Attachment: A screenshot of the page at the moment of detection.

---

📤 Output Example

The Actor produces a clear, structured dataset. Here is a preview of the JSON output:

[
  {
    "status": "MATCH FOUND",
    "url": "https://rockstargames.com/newswire",
    "eventDetected": true,
    "summary": "The article explicitly states that the release date has been set for Fall 2025.",
    "timestamp": "2024-05-20T14:30:00.000Z"
  },
  {
    "status": "NO MATCH",
    "url": "https://ign.com/rumors",
    "eventDetected": false,
    "summary": "The text discusses rumors but explicitly says there is no official confirmation yet.",
    "timestamp": "2024-05-20T14:31:00.000Z"
  }
]

---

⚖️ Cost & Limits

• Apify Platform Usage: This Actor is lightweight. It uses Playwright to render pages (necessary for modern dynamic websites).
• AI API Costs: You must provide your own API keys.
  • Recommendation: Use Groq (Llama 3) or OpenAI (GPT-4o-mini) for the most cost-effective, high-speed monitoring.
  • The Actor truncates text to ~12,000 characters to prevent excessive token usage.

---

Built with ❤️ and </> - Try it now 🚀