Skip to main content
The Product from Website node reads a product or landing page and pulls out the product behind it: name, tagline, description, features, price, target audience, product images, and the logo. Connect its output to generation nodes and they stay faithful to the real product, so you don’t retype product facts or re-upload press images. Under the hood the node loads the page in a headless browser, then a small AI model structures the page content into product facts. Product images and the logo are picked from the images actually found on the page and stored in your workspace.

When to use

  • Seeding a campaign workflow from a live product page
  • Keeping generated copy and visuals faithful to a real product
  • Producing ads, social posts, or videos for many products by feeding a list of URLs
Common patterns: extract_product → generate_text, extract_product → generate_image, extract_product → generate_video.

Inputs

text
Optional. Overrides the Product URL from the node’s configuration. Connect a list of URLs and the node runs once per URL.

Outputs

product
The structured product: name, tagline, description, category, features, price, audience, site colors, and the source URL. Connect it to the Product input on Generate Text, Generate Image, or Generate Video — the product facts are injected into the model’s prompt automatically.
image
Product photos and screenshots found on the page, stored durably in your workspace. Use them as reference images for image or video generation, or edit them directly.
The company or product logo, when one is found on the page.

Configuration

string
The page to read. A bare domain works too: acme.com/widget becomes https://acme.com/widget.
string
Optional guidance appended to the extraction prompt, e.g. “Focus on the enterprise plan” or “Extract the copy in German”.
Extraction reads the page as it is publicly served. Pages behind logins or aggressive bot protection may come back empty or partial. The node costs 2 credits per run.