Output you define
A fixed format keeps field names stable. Bound templates keep working.
AI scraping
AI web scraping that turns public pages into screen content.
The events calendar is already on the website. So is the job board, and the list of opening hours. The Web scraper datasource reads a public page, follows a prompt you write in plain words and returns the fields you asked for as structured data, then refreshes on the schedule you set so that the screens in your lobby change when the page does and nobody has to retype a thing.
How it works
How a scraper datasource works.
Paste a public URL
In Datasources, add a new external datasource and choose Web scraper. The page must be reachable without a login.
Describe what you want
Write a prompt such as list each event with its date, title and image. Name the output language too.
Fix the shape of the data
Let the AI decide, paste example JSON or supply a JSON Schema, so every refresh returns the same fields.
Set the refresh
Pick an interval or write a cron expression. Then bind the data to a slide.
What you can control.
A warning for the wrong format
Paste a schema where the editor expects sample data and it warns you to switch to the JSON Schema option.
Images and links kept
Tell the AI to use only image URLs that exist on the page. The scraper can also keep link URLs from the page.
Wait time
Give the AI more time on long pages. The first run can take a few minutes. Watch its status in the list.
Cached files
Store fetched images locally. Remove broken references, or clear the cache on every refresh.
Rotation and resilience
Shuffle arrays so each refresh shows items in a new order, and ignore brief fetch errors so a blip doesn’t pause it.
Docs and guides.
Show us the page you keep copying.
Send the link before the demo. We’ll build the scraper live and write its prompt with you, and our support team is there if a page layout changes and the output needs a new prompt.