Apollo News Signal Monitor

Apollo News Signal Monitor is an Apify Actor that watches company RSS feeds, blogs, and newsroom pages for fresh signals, scores each item for sales relevance, generates outreach-ready context, and routes the result into Apollo as a task, sequence action, contact update, or review artifact.

Why this Actor exists

Apollo is strong at workflows, tasks, custom fields, and sequence enrollment, but it is not a dedicated external signal monitor for content sources like RSS feeds and newsrooms. This Actor fills that gap by becoming the ingestion layer between public company signals and Apollo execution.

What it does

• Monitors RSS, Atom, blog, and newsroom pages for target companies
• Auto-discovers feed URLs when possible and falls back to HTML extraction when no feed exists
• Stores dedupe state in the default key-value store so the same signal is not processed twice
• Scores items using include and exclude keyword logic plus optional company-specific rules
• Extracts normalized content fields such as title, URL, published date, summary, and a cleaned snippet
• Generates AI-assisted or deterministic icebreakers and outreach framing
• Updates Apollo custom fields, creates review tasks, or enrolls contacts into a sequence
• Emits a dataset of processed signals and a SUMMARY record for downstream automation

Input

Use .actor/input_schema.json or the Apify Console form. The most important fields are:

• sources: array of monitored companies and source URLs
• includeKeywords and excludeKeywords: global relevance controls
• minRelevanceScore: threshold for filtering low-signal items
• mode: detect_only, create_task, enroll_sequence, or human_review
• apolloApiKey: required for Apollo write actions
• apolloSequenceId: required for sequence enrollment
• useAI, llmProvider, and llmApiKey: optional AI personalization settings
• webhookUrl and slackWebhookUrl: optional review notifications
• dedupeWindowDays: how long processed signal fingerprints stay in state

Example input

This configuration matches a typical human-review run against OpenAI, Anthropic, and Stripe news sources:

{
  "dedupeWindowDays": 30,
  "excludeKeywords": [
    "privacy policy",
    "terms",
    "careers",
    "policy",
    "trust center"
  ],
  "fetchFullArticle": true,
  "includeKeywords": [
    "launch",
    "model",
    "API",
    "enterprise",
    "agent",
    "assistant",
    "platform",
    "product",
    "automation",
    "developers"
  ],
  "maxItemsPerRun": 20,
  "minRelevanceScore": 0.45,
  "mode": "human_review",
  "sources": [
    {
      "companyName": "OpenAI",
      "domain": "openai.com",
      "pageUrl": "https://openai.com/news"
    },
    {
      "companyName": "Anthropic",
      "domain": "anthropic.com",
      "pageUrl": "https://www.anthropic.com/news"
    },
    {
      "companyName": "Stripe",
      "domain": "stripe.com",
      "pageUrl": "https://stripe.com/newsroom/news"
    }
  ],
  "useAI": false
}

Source object

Each source supports:

• companyName
• domain
• rssUrl
• pageUrl
• companyIdOptional
• apolloContactIdOptional
• contactEmailOptional
• contactFirstNameOptional
• contactLastNameOptional
• topicRules.includeKeywords
• topicRules.excludeKeywords

Output

Dataset

Each dataset item represents one signal candidate or processed signal. Key fields include:

• signalDetected
• matchedCompany
• matchedContact
• apolloAction
• taskCreated
• sequenceEnrolled
• customFieldsUpdated
• confidence
• generatedIcebreakers
• reviewBrief
• suggestedFirstLine
• outreachAngle
• reviewRequired
• metadata
• errors

Key-value store records

• SUMMARY: run-level metrics and action summary
• STATE: dedupe fingerprints and per-source cursors

Apollo integration notes

The Apollo client implementation is intentionally isolated in ./src/apollo.ts. Apollo API surface details can vary by workspace and endpoint availability, so the Actor:

• favors direct IDs when supplied
• attempts contact lookup with conservative REST calls
• keeps uncertain payload details behind typed helpers and TODO markers
• fails per item instead of failing the whole run

For v1, write actions are best when at least one of these is available:

• apolloContactIdOptional
• contactEmailOptional
• companyIdOptional
• domain or companyName for lookup

Personalization behavior

If useAI is enabled and the configured provider call succeeds, the Actor returns short personalized icebreakers, a signal summary, an outreach angle, and a suggested first line.

If AI is disabled or the provider call fails, the Actor falls back to deterministic templates so runs stay retry-safe and predictable.

Production behavior

• Dedupe is based on a canonical signal hash derived from URL or feed GUID
• Missing publish dates are allowed
• Each item is processed independently with retries around network calls
• Logs stay concise and avoid leaking secrets
• Review mode never auto-enrolls contacts into sequences

Example output

{
  "companyName": "OpenAI",
  "signalType": "rss_post",
  "title": "The next phase of enterprise AI",
  "url": "https://openai.com/index/next-phase-of-enterprise-ai",
  "publishedAt": "2026-04-08T14:00:00.000Z",
  "sourceName": "OpenAI News",
  "summary": "OpenAI outlines the next phase of enterprise AI, as adoption accelerates across industries with Frontier, ChatGPT Enterprise, Codex, and company-wide AI agents.",
  "snippet": "OpenAI outlines the next phase of enterprise AI, as adoption accelerates across industries with Frontier, ChatGPT Enterprise, Codex, and company-wide AI agents.",
  "relevanceScore": 0.65,
  "confidence": 0.7,
  "signalSummary": "OpenAI published a fresh signal around AI product momentum.",
  "reviewBrief": "OpenAI: The next phase of enterprise AI\nScore: 0.65\nSummary: OpenAI published a fresh signal around AI product momentum.\nSuggested opener: Noticed OpenAI's recent update on AI product momentum and thought it was a strong signal for current priorities.",
  "outreachAngle": "Use the update as a timely hook to connect your solution to AI product momentum outcomes and near-term execution priorities.",
  "suggestedFirstLine": "Noticed OpenAI's recent update on AI product momentum and thought it was a strong signal for current priorities.",
  "icebreakers": [
    "Saw OpenAI's latest post on AI product momentum and the timing feels especially relevant.",
    "Your recent update about AI product momentum stood out, particularly the practical execution angle."
  ],
  "generatedIcebreakers": [
    "Saw OpenAI's latest post on AI product momentum and the timing feels especially relevant.",
    "Your recent update about AI product momentum stood out, particularly the practical execution angle."
  ],
  "signalDetected": true,
  "matchedCompany": false,
  "matchedContact": false,
  "apolloAction": "human_review",
  "taskCreated": false,
  "sequenceEnrolled": false,
  "customFieldsUpdated": false,
  "reviewRequired": true,
  "errors": [],
  "metadata": {
    "sourceFeedUrl": "https://openai.com/blog/rss.xml",
    "personalizationProvider": "template",
    "fallbackUsed": true
  },
  "status": "processed"
}

Future extensions

• Competitor monitoring across peer company feeds and launch pages
• Hiring and organizational-news signals for trigger-based prospecting
• Negative signal and churn monitoring based on incidents, layoffs, and pricing changes
• Social listening extensions for supported third-party sources