The agent interacts with websites by perfoming actions such as clicking, typing and scrolling. It can therefore perform operations of any complexity.

An ‘action’ step is one that requires the agent to interact with the page by performing one or more actions. You define what you want to do and the agent decides which actions to perform to accomplish the task.

Action steps are very flexible - they don’t break if your UI changes, unexpected pop-ups appear, or the necessary element is beyond the current visibility of the agent. You can be as precise as you want in defining action steps.

​
Examples of Action Steps

The way you define action steps allows you to balance flexibility and granularity to suit every unique use case.

Here are some simple examples requiring one or a few actions:

“Click on ‘Sign In’” [a simple step, likely requiring only one action]

“Go to pricing” [the agent assesses how it should do this, which likely involves clicking on the ‘Pricing’ tab]

“Type the name ‘Richard Hendricks’” [requires two actions - clicking on the ‘Name’ field and typing the name]

“Submit the form” [this may require scrolling down to the ‘Submit’ button]

“Close the popup” [requires identifying where the ‘close’ button is]

Here are some more complex examples that require multiple actions:

“Scroll down until you see the Testimonials section”

“Fill out the sign-up form”

“Select travel dates for 28 May to 3 June”

“Search for 2-bedroom apartments in London”