Browser Node

The Browser node is a powerful tool that enables browser automation. It can handle various tasks such as gathering information from websites, data entry, and downloading.

This tutorial introduces a “workflow for researching the company that operates SamuraiAI.”

Tutorial

Browser Node

First, create a new workflow in Studio.

Add a Browser node and copy and paste the following content into the node’s prompt:

1. go to https://samuraix.ai/
2. Find the link to the operating company and navigate to the page
3. Search for detailed information about the operating company and output it

Data Extraction

Next, add a new LLM node to the right of the Browser node and copy and paste the following content:

Output only the company address from the following message.

{browser-1.response}

Overall Workflow

The final workflow looks like this:

Running the Workflow

Now you’re ready! Click the debug button to run the workflow.

When executed, the browser will automatically open and collect company information from the specified website. Finally, the address information of SamuraiAI’s operating company was output 🚀

Note

When running browser operations for the first time on Windows, you may be asked for network permissions, so please grant them.

Browser operations are designed to work at high speed, and you can see that they operate the browser at the same speed as or faster than a human (video playback speed: 1.0x).

Interestingly, the AI automatically extracts the operating company’s URL from just the service site URL provided and navigates further to the corporate information page. There’s no need for humans to specify which HTML elements are needed.

Customizing the Process

Let’s now customize the workflow. This time, we’ll change the service URL from “https://samuraix.ai/” to “https://slack.com/intl/ja-jp/”.

After making the change, click “Save” and run it again.

Tip

The first time, you may see a browser popup asking if you want to open the desktop app. In that case, select No so the browser processing continues. In SamuraiAI, there’s no problem with humans actively intervening in the process. If the AI is having trouble, help it out.

With just one line change, we were able to retrieve Slack’s operating company. Since SamuraiAI automatically operates the browser, humans can immediately check the output data and perform additional searches if needed.

Learn More

Recommended Environment

• The Browser node is only available in the desktop version of SamuraiAI. It is not available in the web version.
• A stable network connection is recommended. If the network is unstable, browser operations may fail midway.
• The higher the PC specifications, the smoother the browser operations. Memory and CPU performance are particularly important.

Browser Node Features

The Browser node launches a browser dedicated to SamuraiAI on your local device and performs operations. This provides the following features:

• Secure Authentication: Website login information is stored within your PC and is never sent externally.
• Integration with Local Data: You can upload files from your PC or check downloaded files on your PC.
• High-Speed Operation: Automation processing optimized for browser operations enables high-speed performance.
• Easy User Intervention: Users can intervene during browser operations to assist manually.

Recommended Authentication Method

When automating login operations with the Browser node in SamuraiAI, we recommend the following authentication method:

1. Execute any browser operation workflow to launch the browser.
2. Manually perform login operations. This caches the authentication information within your local device, so you won’t need to log in again until the authentication expires.

File Downloads

Browser operations can also download files from websites. Downloaded files are saved in the following folder:

“Menu Bar” -> “File” -> “Open Download Folder”

Websites That Are Difficult to Automate

SamuraiAI analyzes a website’s accessibility tree to perform operations. This significantly reduces token consumption and enables high-speed operations. On the other hand, operations may not work well on websites that lack sufficient accessibility considerations.

Also, if there is an extremely large amount of information on a single page, LLM token limitations may prevent correct operations.

Operation Support

Operation	Support	Details
Site Information Retrieval	✅️
Text Input	✅️
Authentication such as Login	✅️
File Download	✅️
File Upload	✅️
Screenshot Capture	✅️
reCAPTCHA	△	Simple ones only
Bot Protection Bypass	△	Human intervention required