Google AI Overview Scraper

Extract the AI Overview that Google visibly renders for search queries, including overview text, bullets, and supporting source links. This version uses direct DOM extraction only, so it does not call an external LLM provider and does not require an OpenRouter API key.

This actor is an extraction tool, not an answer-generation tool. If Google does not show an AI Overview for a query, the actor returns overviewPresent: false and a warning. It never creates a replacement answer.

What it returns

The dataset contains one item per processed query:

Field	Description
query	The Google Search query sent by the actor.
searchUrl	The rendered Google Search URL used as the source page.
overviewPresent	Whether a Google AI Overview was detected.
overviewText	Visible overview text, or null when no overview was available.
overviewBullets	Short bullets found in the visible overview.
sources	Supporting links captured from the rendered Google page.
extractor	dom when an overview was detected, or none.
warnings	Non-fatal details such as no overview or a Google challenge page.
error	A recoverable query error, or null.
scrapedAt	ISO 8601 collection timestamp.

The SUMMARY key-value-store record contains request totals, locale, proxy mode, query price, and run-level warnings.

When to use this actor

Use it to observe Google AI Overview coverage for a list of queries, compare visibility across country/language settings, collect the sources Google exposes alongside an overview, or feed structured visibility evidence into an SEO or research pipeline.

Do not use it to ask an AI model a question, generate a synthetic answer, scrape Google account data, or bypass a Google challenge. For standard organic result extraction without AI Overview detection, use a Google Search or SERP-focused actor.

Input

{
  "queries": [
    "what is retrieval augmented generation",
    "how does a heat pump work"
  ],
  "countryCode": "us",
  "languageCode": "en",
  "useApifyProxy": false
}

• queries — one natural-language search query per line; up to 25 queries per run.
• countryCode — two-letter Google country parameter such as us, gb, or pk; defaults to us.
• languageCode — two-letter Google language parameter such as en, de, or ur; defaults to en.
• useApifyProxy — defaults to false to avoid proxy charges. Enable it when Google challenges the default cloud IP; proxy usage may add Apify proxy charges.

Output example

{
  "query": "what is retrieval augmented generation",
  "status": "success",
  "overviewPresent": true,
  "overviewText": "Retrieval-augmented generation combines information retrieval with text generation...",
  "overviewBullets": [
    "Retrieve relevant documents",
    "Ground the generated response in those documents"
  ],
  "sources": [
    {
      "title": "Retrieval-augmented generation",
      "url": "https://example.com/rag",
      "domain": "example.com",
      "snippet": "Background on RAG"
    }
  ],
  "extractor": "dom",
  "warnings": [],
  "error": null,
  "scrapedAt": "2026-08-08T12:00:00.000Z"
}

If an eligible query produces no overview, the item remains a truthful observation with overviewPresent: false, overviewText: null, an empty overviewBullets array, and a warning explaining that Google did not expose a detectable overview. Google availability varies by country, language, account, query, device, and time.

Pricing

The included PPE definition charges:

Event	Price	When it fires
apify-actor-start	$0.00005	Once per run.
google-ai-overview-result	$0.005	Once per valid query processed, including no-overview results.

A five-query run has a maximum event charge of $0.02505 before any Apify plan or proxy charges. The actor prints this cap at run start and stops additional work if the PPE event limit is reached.

Reliability and limitations

The actor uses a single Playwright browser session with low concurrency and retries temporary navigation failures. It extracts only text and links visible in the rendered Google page; there is no LLM cleanup or answer synthesis. Google changes its markup and may show different features for different sessions, so a missing overview means only that this run did not detect one.

Google challenge pages are reported as errors or warnings. The actor does not attempt to defeat CAPTCHA, login walls, or access controls. Enable Apify Proxy only when needed, because proxy traffic may create additional usage charges.

Use the returned searchUrl and source links for review, and independently verify important claims. This actor does not provide legal, medical, financial, or other professional advice.

API and MCP usage

Start the actor through the Apify API with an input object matching INPUT_SCHEMA.json. The dataset output is available from the run’s default dataset, and the SUMMARY record is available from the run’s default key-value store. The flat, explicit output fields and stable null values are intended for Apify MCP and other agent clients.

Legal and responsible use

Use this actor only where you have a lawful basis and permission to access the target service. Respect Google’s Terms of Service, robots and access policies, applicable privacy law, and any contractual restrictions. Do not submit personal, confidential, or sensitive queries.

References

• Google Search Help: AI Overviews