Scrape a website by saying what you want

Q: Can it handle cookie banners?

Yes, `handle_cookie_consent` is on by default.

Name the data you need in plain language, such as product name and product price, and the scraper uses Google's Gemini to find those values on the page. No selectors to write.

Start AI scraping on the Apify Store · Version 0.0.52, updated 2026-07-22

Who it is for

Marketers, researchers and other non-developers who need data from websites without writing a scraper.

What you get

Plain-language fields

List the items you want, separated by commas, and the Actor finds and extracts them.

Dynamic pages

Handles cookie consent, infinite scroll, pagination and links to follow.

Desktop or mobile

Scrape as a desktop browser or as a mobile device such as an iPhone or Samsung Galaxy.

Items or chunks

Output structured items, or text chunks for use in AI pipelines.

How it works

Put the pages in start_urls and describe what you want in instructions or output_fields.
Optionally turn on follow_links, set max_pages or a pagination_selector.
Choose a gemini_model and start the run.

Also runs from AI assistants through Apify's MCP server.

Use cases

Collecting product names and prices from a webshop.
Pulling contact details from a list of company websites.
Turning web pages into chunks for a retrieval pipeline.

Pricing

Startup costs Startup costs for starting the actor	$0.02
Vision call to LLM Vision Call to LLM for processing image of resolution 1920x1080	$0.01

You pay per event listed above; there is no monthly fee.

Inputs

`start_urls` required	Start URLs
`instructions`	Instructions
`output_mode`	Output mode `items` · `chunks` default: `items`
`chunk_size`	Chunk size (tokens) default: `512`
`chunk_overlap`	Chunk overlap (tokens) default: `64`
`output_fields`	Output fields (optional)
`follow_links`	Follow links to detail pages default: `False`
`max_followed_links`	Max followed links default: `10`
`max_pages`	Max pages (pagination) default: `1`
`pagination_selector`	Pagination selector (optional)
`proxyConfiguration`	Proxy configuration
`has_infinite_scroll`	Has infinite scroll default: `False`
`above_fold_only`	Above fold only default: `False`
`save_screenshots`	Save screenshots default: `False`
`handle_cookie_consent`	Handle cookie consent default: `True`
`gemini_model`	Gemini model `gemini-3.1-flash-lite` · `gemini-2.5-flash` · `gemini-2.5-pro` default: `gemini-3.1-flash-lite`
`device_type`	Device type `desktop` · `mobile` default: `desktop`
`mobile_device_model`	Mobile device model `iPhone 15 Pro Max` · `iPhone 15 Pro` · `iPhone 15 Plus` · `iPhone 15` · `iPhone SE (2022)` · `Samsung Galaxy S24` · `Samsung Galaxy S23` · `Custom` default: `iPhone 15`
`custom_viewport_width`	Custom viewport width default: `393`
`custom_viewport_height`	Custom viewport height default: `852`
`max_concurrent_screenshots`	Max concurrent screenshots default: `4`
`screenshot_timeout`	Screenshot timeout default: `60`

Limits

Still experimental: it relies on vision, so results can be inconsistent.

Questions

What does it cost?

$0.01 per vision call plus $0.02 per start.

Do I need to know CSS selectors?

No. You describe the fields in plain language.

Can it handle cookie banners?

Yes, handle_cookie_consent is on by default.

Start AI scraping