Visual changes made by an assistant

An assistant can add, edit and remove visual changes the way the visual editor does. Here is how they appear in the editor, and what it cannot change.

2 min read


An assistant with Build experiments can add, edit and remove the same kinds of visual change the editor makes: text, HTML, images, background images, styles (including hover styles), link addresses and other attributes, hiding an element and moving one. It works on Draft and QA experiments only.

How it finds the element

Pertento never loads your pages itself. The assistant looks at your page with its own browser tools, for example a browser extension, picks the element and writes the change. Then you check it on your real site with a QA preview.

Check that each change found its element

  1. 1Ask the assistant to start QA and give you the preview link for the variant.
  2. 2Open the link in your browser.
  3. 3Ask the assistant to read the preview report.

The report says which changes found their element on the page and which found nothing, so the assistant can fix a change that missed. See Checking a variant on your live site with QA.

In the editor

An assistant's changes appear in the variant's change list like any other, titled with the assistant's short description and marked Added by an agent, with whose token added it and when. You can select one and fine-tune it in the editor. It stays marked as the agent's after you edit it.

What an assistant cannot do

  • Add HTML that runs code. Code belongs in the variant's JavaScript.
  • Target the whole page (the page body or root) with any change.
  • Put page data into an image or link address, or point an image address at anything but an image, video or audio element, or a link address at anything but a link.
  • Edit a change that the editor aimed at several elements at once. It removes that change and adds its own instead.