Skip to main content
The best way to master Simular Pro is to learn from the example instructions (prompts) and corresponding Simulang on this page. Just run the instructions on your Simular Pro to reproduce similar Simulang code, or copy the Simulang code on this page and directly execute. Find out how to: You can always refer to the SimuLang API documentation if you need a refresher on each Simulang action.
Most of the Simulang code on this page were generated in one shot from the instructions. Since there may be differences in individual machine configurations (e.g. browser settings), we recommend that you double check the Simulang before executing.
If Simular Pro did not click on an element you intended, try these various methods for element localization.

Reading structured information

Many workflows start with extracting information from an application or webpage. Moreover, the extracted information often needs to be represented in a particular structure. For example:
  • A house on sale has the following fields: address, square footage, number of beds and baths, etc.
  • An email has these fields: sender address, time sent, title, etc.
This section shows you how to instruct Simular Pro to extract information from diverse sources into structured forms. The resulting information will be ready for downstream workflow steps, such as iterating a subtask over multiple elements or writing structured outputs to files.

Community and communication applications: Extracting channel names on a desktop community app

Open Discord
Get the page content and extract the top 10 server names visible in the left sidebar.
Do not include direct messages.

Shopping: Get all store locations of a grocery chain

Go to Wegmans stores webpage.
Get all the locaitons and print them out.

Social media URL scraping: Get a company’s social media URLs

Go to simular.ai
Get all of their social media URLs and print them out

News: Get all headlines from a news site

Go to techcrunch, get all the news headlines, and print them out.

Investing: Get an investment firm’s list of portfolio companies

Go to a16z's portfolio page and get all the features company names on the page.

Email management: Get the sender, date, and summary of the first unread email.

Go to gmail.com
Click the first unread email
From the email content, get: sender email address, date, summary of topic

Real estate: Get address, number of bedrooms and bathrooms, square footage, and price of homes in Palo Alto

Go to redfin palo alto zip 14325
Get all the address, number of bed, number of bath, square feet, and price of the houses

Repeating a task on multiple objects on a page

Having successfully used Simular Pro to read structured information, such as getting
  1. all the names of companies on a page, or
  2. all the store locations of a grocery chain, or
  3. all the influencer channel names, etc.
now you may need to repeat some task on each of those items, such as:
  1. checking each company’s stock prices,
  2. each grocery store’s items on sale
  3. each influencer’s subscriber count
The following examples show how to instruct Simular Pro to iterate through all the items and do a follow-up task reliably on each item.

Opening and reading all Simular blog posts

Go to simular.ai/blog
Get all the blog post URLs
For each blog post URL, open it, extract the post title and a short summary, then close the tab.
Print out the post titles and summaries.

Structured information output

The sections above show you how Simular Pro can read structured information and repeats a task on 100 webpages - now you need it to write that valuable information somewhere. This section shows you many options to write structured output to a local file on disk or a web browser application. If you cannot find an output format you need, just ask us on Discord.

Writing information into a Google Sheet

Open a new Google Sheet
Add headers Name, Summary, Date
Write "Simular", "Simular pro can do anything!", "08/30/2025" into the first row.

Writing information into Excel

Open a new blank sheet in Excel
Add headers Name, Summary, Date
Write "Simular", "Simular pro can do anything!", "08/30/2025" into the first row.

Clicking on UI elements

Sometimes, the basic click({at: <description of some element>}) may not be enough for Simular Pro to identify a unique element. This may happen when more than one element share the same text description, or when the target element only has a visual label (e.g., image or icon). In these cases, you have two options:
  1. In Settings, enable vision fallback for the click action. Simular Pro will automatically attempt to use the best method for UI localization.
  2. Use more precise inputs to the click function, as shown by examples below. You can find the full specification of inputs to the click function here.
Use the element picker tool to check the text description of your target element.

Clicking one of multiple elements that have the same description or image.

This pattern is a powerful and fast way to localize a target element whose textual description may not be unique:
click({
    at: <something not unique>, 
    spatialRelation: <a comma-separated string of spatial relationships>, 
    anchorConcept: <something unique>
})
You may use one from each of the following three spatial relation categories for spatialRelation:
  • distance: closest, furthest
  • positional filters: left, right, above, below,
  • containment filters: contains, containedIn
For example, this will find the unique element called “link Introduction” that is above of and closest to a unique element called “heading Simular Pro”:
click({
    at: "link Introduction", 
    spatialRelation: "above,closest", 
    anchorConcept: "heading Simular Pro"
})
Here is an example drawn from this current page:
Go to docs.simular.ai
Click on Introduction closest to Getting Started
Click on Introduction closest to Simular Browser
Click on Introduction closest to Agent S
Click on Introduction closest to Simular Pro
When attempting to close a popup window, you may need to distinguish between the close button of the popup window and the close button of the outer application. You can instruct Simular Pro to click the close button contained in the dialog window.
Use the element picker tool to get the description of the anchor element.
Click the close button contained in the group dialog.

Clicking on an element without text description

If your target element is a pure image without any text description, then you should use the vision mode to localize the element:
click({at: <describe your element>, mode: "vision"})
Use the element picker tool to check whether your target element has any useful text description.
Go to docs.simular.ai
Click the Simular logo using vision
Sometimes you may need to click on an empty space on the current screen, such as when applying a signature to some area of a document.
Click the empty space below the date.

Clicking on an element using both text and screenshot

If neither specifying spatial relations nor pure vision work, you may need a combination of both to localize an element. The mode textAndScreenshot is able to localize any element that has a text description.
click({at: <description of element>, mode: "textAndScreenshot"})
You may use any natural language for the description of the element, including any referring expressions (e.g., “link Introduction in the Simular Browser section”)
Open docs.simular.ai
Click the introduction link in the simular browser section