Patents Scraper — Google Patents Search, Prior Art, Citations

Patent and prior-art search over Google Patents. Search by full text, assignee, inventor, patent office, granted or application, utility or design, language, and a date range on the priority, filing or publication date — and get one row per patent: publication number, title, snippet, assignee, inventor, all four dates, family offices, drawing thumbnail and a direct PDF link.

Turn on Open each patent page and every row gains the parts a search result leaves out: the full abstract, every inventor, CPC classifications, legal status, the patents it cites and how many patents cite it.

No API key and no account. One search request returns 100 patents, so a 1,000-patent sweep is ten requests and about twenty seconds.

Why this Actor

• A whole portfolio in one run. An assignee plus a date range returns everything a company has filed since, newest first, 100 patents per request — no paging through a website.
• Citations both ways. With detail pages on, each row carries the patents it cites and how many patents cite it, which is the usual shorthand for how much a filing mattered.
• Classifications you can group by. CPC codes come back as full subgroups like A47K3/022, with the hierarchy above them stripped out, so a pivot table works straight away.
• No key, no account, no proxy — and the search request is the same one the website makes, so any query that works there works here.

Sample output

One item of the dataset (long values shortened):

See the Output tab of any run.

What you get

• A portfolio in one run: put a company in Assignees and a date in From date and you have everything it has filed since, sorted newest first
• Prior-art lists with the snippet that matched, the PDF link and the family size next to each hit
• Citations in both directions with the detail pages on: what a patent cites, and how many later patents cite it — the usual proxy for how important it turned out to be
• CPC classifications as a clean list of full subgroups (A47K3/022), not the hierarchy above them
• Legal status as the registry words it — Active, Expired - Fee Related, Withdrawn
• Filters that actually narrow: patent office, granted versus application, design patents, language, and which of the three dates the range applies to

Google serves at most 1,000 results for a single query, so a wide sweep is best split by year, country or assignee — the run says so in the log when it hits that ceiling.

Input example

{
  "queries": [
    "solid state battery"
  ],
  "dateField": "priority",
  "sortBy": "relevance",
  "includeDetails": false,
  "concurrency": 4,
  "maxResults": 100
}

How much does it cost?

Pay per patent — no subscription, no minimum, no charge for platform usage.

Volume	Price
1,000 patents	$1.00 (+ $2.00 with detail)
10,000 patents	$10.00 (+ $20.00 with detail)
100,000 patents	$100.00 (+ $200.00 with detail)

The Apify free plan includes $5 of usage every month — about 5,000 patents with this Actor, no card needed. Nothing else is charged: platform usage is included in the price, and Apify Bronze, Silver and Gold subscribers get 10%, 20% and 30% off these prices.

Use it from code, n8n, Make, Zapier or an AI agent

Run the Actor and download the dataset in one call (JSON by default; add &format=csv or xlsx):

curl -X POST "https://api.apify.com/v2/acts/chorelet~patents-scraper/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"queries": ["solid state battery"], "dateField": "priority", "sortBy": "relevance", "includeDetails": false, "concurrency": 4, "maxResults": 100}'

Python:

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("chorelet/patents-scraper").call(run_input={"queries": ["solid state battery"], "dateField": "priority", "sortBy": "relevance", "includeDetails": false, "concurrency": 4, "maxResults": 100})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

• n8n, Make, Zapier — use the Apify node/module: run the Actor, then "get dataset items".
• Google Sheets, Slack, webhooks — add an integration on the run's Integrations tab.
• AI agents — the Actor is available as a tool through the Apify MCP server; the dataset schema describes every field for the model.
• Schedules — run it hourly, daily or weekly from the Schedules tab.

FAQ

How many patents can one search return?

Google serves at most 1,000 results for a single query. The run logs it when a search is wider than that; split by year, country or assignee to go deeper.

What do the detail pages add?

The full abstract, every inventor, CPC classifications, legal status, the list of patents cited and the number of patents citing it. They cost one request per patent, which is why they are a separate charge.

Can I search by company?

Yes — Assignees matches the owner on the patent. Add a date range on the priority or filing date to get just the recent filings.

Which patent offices are covered?

Everything Google indexes: US, EP, WO, CN, JP, KR, DE and over a hundred more. Leave Patent offices empty for all of them, or list the two-letter codes you want.

Are applications included?

Yes, and you can filter: Granted returns issued patents, Application returns published applications, empty returns both.

Is the data free to reuse?

Patent documents themselves are public records. This Actor reads the public search endpoint and the public patent pages; it adds no licence of its own.

Support

Questions, missing fields or a source that changed? Open an issue on the Issues tab or write to support@chorelet.app — problems are usually fixed within a day, and the Actor is checked every morning by an automated test run. If the Actor saved you time, a short review on its Store page helps other people find it.