How to use Multion?

Leon Chase

• 4 min read

Multion is a tool designed to help developers and users automate interactions with web applications using AI. It allows you to control web browsers programmatically, interact with websites, and perform tasks like filling out forms, clicking buttons, navigating pages, and extracting data. Multion leverages AI to understand and interact with web pages in a more human-like way, making it easier to automate complex workflows without needing to write extensive code.

Here’s a step-by-step guide on how to use Multion:

1. Setting Up Multion

Step 1: Install Multion

  • Python Installation: Multion is typically used as a Python library. You can install it using pip:
    pip install multion

Step 2: API Key

  • API Key: After installing Multion, you’ll need an API key to authenticate your requests. You can obtain this by signing up on the Multion website.

    Once you have the API key, set it as an environment variable or pass it directly in your code:

    import os
os.environ['MULTION_API_KEY'] = 'your_api_key_here'

2. Basic Usage of Multion

Step 3: Import and Initialize Multion

You can start using Multion by importing it into your Python script and initializing the client:

from multion import Multion

# Initialize Multion client
multion = Multion()

Step 4: Start a New Session

To interact with a website, you first need to start a new browser session:

session = multion.new_session(url="https://example.com")

This will open a new browser session at the specified URL (https://example.com in this case).

3. Interacting with Websites

Step 5: Perform Actions on the Web Page

Once the session is started, you can instruct Multion to perform various actions on the web page. For example, you can fill out forms, click buttons, or navigate between pages.

Example 1: Filling Out a Form

You can use natural language commands to tell Multion what to do. For example, if you want to fill out a login form:

response = multion.execute_command(session_id=session['session_id'], command="Login with username 'testuser' and password 'mypassword'")

Multion will use AI to locate the appropriate fields on the page and fill them out accordingly.

Example 2: Clicking a Button

If you want to click a button (e.g., "Submit" or "Next"), you can issue a command like:

response = multion.execute_command(session_id=session['session_id'], command="Click the 'Submit' button")
Example 3: Navigating to Another Page

You can also navigate to another page within the same session:

response = multion.execute_command(session_id=session['session_id'], command="Go to https://example.com/dashboard")

4. Extracting Data from Websites

Step 6: Extracting Information

After performing actions on a website, you may want to extract specific information. You can use Multion to extract data based on natural language queries.

Example: Extracting Product Prices

If you're on an e-commerce site and want to extract product prices:

response = multion.execute_command(session_id=session['session_id'], command="Extract all product names and prices from the page")
print(response['data'])

Multion will return the extracted data in a structured format (e.g., JSON), which you can then process further.

5. Closing the Session

Step 7: End the Session

Once you’re done interacting with the website, you should close the session to free up resources:

multion.close_session(session_id=session['session_id'])

6. Advanced Features

Handling Dynamic Content

Multion is capable of handling dynamic content rendered by JavaScript. This means that even if a website loads content asynchronously (e.g., via AJAX), Multion can still interact with and extract data from it.

Customizing Commands

You can customize the commands you send to Multion to perform more complex tasks. For example, you can chain multiple actions together:

commands = [
    "Navigate to https://example.com/login",
    "Login with username 'testuser' and password 'mypassword'",
    "Go to https://example.com/dashboard",
    "Extract all recent notifications"
]

for cmd in commands:
    response = multion.execute_command(session_id=session['session_id'], command=cmd)
    print(response['data'])

Error Handling

Multion provides error handling for cases where the AI might not be able to interpret a command correctly. You can check the response for errors and handle them accordingly:

if 'error' in response:
    print(f"Error: {response['error']}")
else:
    print("Command executed successfully!")

7. Use Cases for Multion

E-commerce Automation

  • Automate tasks like logging into e-commerce platforms, searching for products, adding items to carts, and checking out.
  • Extract product details, prices, and reviews for market analysis.

Form Automation

  • Automatically fill out and submit forms on websites (e.g., registration forms, surveys, etc.).
  • Automate repetitive tasks like submitting job applications or filling out lead generation forms.

Web Scraping

  • Extract structured data from websites without needing to manually parse HTML or deal with complex selectors.
  • Scrape dynamic content that changes frequently or is loaded via JavaScript.

Testing and QA

  • Automate user interactions for testing web applications (e.g., simulating user behavior like clicking buttons, filling forms, and navigating pages).
  • Perform regression testing by automating common user workflows.

Chatbots and Virtual Assistants

  • Integrate Multion with chatbots to allow users to interact with websites through natural language commands (e.g., "Check my bank balance" or "Book a flight").

8. Pros and Cons of Using Multion

Pros:

  • Ease of Use: Multion abstracts away much of the complexity of traditional web scraping and automation tools by allowing users to interact with websites using natural language commands.
  • AI-Powered: The AI understands the structure of web pages and can perform tasks like filling out forms, clicking buttons, and extracting data without requiring manual coding.
  • Handles Dynamic Content: Multion can handle modern, JavaScript-heavy websites, ensuring that dynamically loaded content is captured accurately.
  • Cross-Platform: Works across different websites and platforms without needing to write custom scripts for each site.

Cons:

  • Cost: Multion is a paid service, and depending on your usage, it could become expensive for large-scale automation tasks.
  • Limited Customization: While Multion simplifies many tasks, it may not offer the same level of fine-grained control as traditional scraping tools for highly complex or custom workflows.
  • Dependence on AI: Since Multion relies on AI to interpret commands, there may be occasional inaccuracies or misunderstandings, especially with poorly structured websites.

Conclusion

Multion is a powerful tool for automating interactions with web applications using AI. It simplifies tasks like filling out forms, clicking buttons, navigating pages, and extracting data by allowing users to issue natural language commands. This makes it accessible to both technical and non-technical users, and it’s particularly useful for automating repetitive tasks, web scraping, and testing web applications.

However, while Multion offers ease of use and AI-powered automation, it may not be as customizable or cost-effective for very large-scale or highly specialized tasks compared to traditional scraping tools like BeautifulSoup, Selenium, or Scrapy.

If you're looking for a quick and easy way to automate web interactions without diving deep into code, Multion is a great choice.