AI Brand Visibility Monitor — Track Your Brand in ChatGPT, Gemini & Perplexity

This AI brand visibility monitor is a GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) tool that tracks whether your brand appears in AI-generated answers — the new search results. It runs your buying-intent prompts (e.g. "What is the best note-taking app for teams?") against ChatGPT (OpenAI), Google Gemini and Perplexity using your own API keys, then analyzes every answer: is your brand mentioned, in what position vs competitors, how often, and with what sentiment. Perfect for AI SEO audits, LLM visibility tracking, share-of-voice in AI answers, and competitor benchmarking in generative search.

Why monitor AI answers?

AI answers are the new search results. A growing share of product discovery happens inside ChatGPT, Gemini and Perplexity instead of Google's ten blue links. If the AI doesn't mention your brand when a user asks about your category, you are invisible to that buyer — no ranking, no click, no chance.

Dedicated AI-visibility SaaS tools charge $300+/month for this. This actor is pay-per-check and uses your own LLM API keys, so a full weekly audit of 20 prompts across 3 providers costs you cents of API usage plus a tiny per-check fee. No subscription, no lock-in.

What you get for each prompt × provider check

{
  "prompt": "What is the best note-taking app for teams?",
  "provider": "openai",
  "model": "gpt-5",
  "brand": "Notion",
  "mentioned": true,
  "mentionPosition": 1,
  "firstMentionCharIndex": 0,
  "mentionCount": 2,
  "competitorsMentioned": [
    { "name": "Evernote", "mentionCount": 1, "firstMentionCharIndex": 94 },
    { "name": "Obsidian", "mentionCount": 1, "firstMentionCharIndex": 134 }
  ],
  "sentimentHint": "positive",
  "answerExcerpt": "Notion is widely considered the best note-taking app for teams…",
  "changed": false,
  "checkedAt": "2026-07-02T09:30:00.000Z"
}

• mentionPosition — 1-based rank of your brand among all tracked brands (yours + competitors) by first occurrence in the answer. Position 1 = the AI names you first.
• sentimentHint — a simple keyword heuristic (positive words like best, recommended, leading vs negative words like avoid, expensive, lacks near your brand mentions). It is a hint, not an NLP sentiment model.
• changed — only in monitor mode: true when visibility flipped (gained or lost) vs the previous run for that prompt+provider.
• fullAnswer — included only when you enable includeFullAnswer.

How to use it

1. Set your brand (e.g. Notion) and optionally competitors (e.g. Evernote, Obsidian).
2. Add prompts — the real questions your buyers ask AI assistants.
3. Add your API keys under providers — any subset of:

   {
     "openai":     { "apiKey": "sk-...",   "model": "gpt-5" },
     "gemini":     { "apiKey": "AIza...",  "model": "gemini-2.5-flash" },
     "perplexity": { "apiKey": "pplx-...", "model": "sonar" }
   }
   

   Only providers with an apiKey are queried; the rest are skipped.
4. (Optional) Enable monitor mode and schedule the actor (e.g. weekly). Each run compares against the previous one and flags visibility changes — combine with a webhook for instant alerts.

Why bring your own keys (BYO)? You control the model, the rate limits and the cost — you pay your providers directly at API prices instead of a marked-up SaaS bundle. Your keys are entered in a secret input field (encrypted by Apify), used only to call the providers you configured, and never stored or logged by the actor.

Use cases

• Agency GEO reports: run a client's prompt set weekly and deliver an AI-visibility report (mentioned/not, position, sentiment) across ChatGPT, Gemini and Perplexity.
• Brand launch tracking: watch your new brand start appearing in AI answers as your content and PR footprint grows.
• Competitor benchmarking: see exactly which competitors the AIs recommend first for your category prompts — and how you compare.
• Visibility-loss alerting: monitor mode + a schedule + a webhook = an alert the moment an AI stops (or starts) mentioning you.

Pricing

Pay per prompt-check — one small charge per (prompt × provider) result delivered. Failed provider calls and configuration errors are not charged. LLM token costs are billed by your providers directly at their API rates.

FAQ

Are my API keys safe?

Yes. The providers input field is marked secret, so Apify stores it encrypted. The actor sends each key only to its own provider's official API endpoint, and keys are never written to logs, dataset items or error messages.

Which models are supported?

Any chat model your key can access: OpenAI chat-completions models (default gpt-5), Google Gemini generateContent models (default gemini-2.5-flash), and Perplexity chat models (default sonar). Set the model field per provider to override.

How is sentiment computed?

It's a deliberately simple keyword heuristic: the actor looks at ±200 characters around each brand mention and counts positive words (best, recommended, popular, powerful, leading, excellent, great) vs negative words (avoid, worse, expensive, limited, lacks, poor). It gives a fast directional signal, not a full sentiment analysis — treat sentimentHint accordingly.

What happens if I run it without any API keys?

The run finishes gracefully: it logs a warning ("no providers configured") and pushes a single NO_PROVIDERS_CONFIGURED item (not charged) so schedules and integrations can detect the misconfiguration.

How do I schedule a weekly check?

In the Apify Console, create a Schedule (e.g. every Monday 08:00), attach this actor with your saved input and monitorMode: true, and optionally add a webhook on run success to push results (including changed flags) to Slack, email or your reporting pipeline.

Can AI agents use this actor?

Yes — flat input, clean JSON output, and like every Apify actor it is consumable via the Apify MCP server, so agents can run visibility checks as a tool.