Brainz Lab / Products
Building Now

Vision

Browser Automation

Give Claude eyes and hands in the browser. AI sees your app like a user, clicks, types, and tests—automatically.

Before

How Claude sees your app today

C
Claude
Y

You

Can you test if our signup flow is working correctly?

C

Claude

I'd like to help, but I can't interact with your live application.

I can't open a browser, fill in forms, click buttons, or verify what users see.

Could you walk through the signup flow and describe what happens?

Y

You

*Opens browser, starts screen recording, manually tests each step...*

After

How Claude sees your app with Vision

C
Claude
Vision connected
Y

You

Can you test if our signup flow is working correctly?

C

Claude

I'll test the signup flow now...

Using vision.navigate

vision.navigate("https://app.example.com/signup")

Screenshot: Signup page loaded

Signup page loaded. I see email and password fields. Let me fill them in...

Using vision.type

vision.type("#email", "[email protected]")

vision.type("#password", "SecurePass123!")

vision.click("#signup-button")

Screenshot: "Welcome! Check your email"

Signup works. Form submitted successfully and redirected to welcome page.

I noticed the password field doesn't show strength requirements. Want me to test edge cases?

Test time: 12 seconds

The Difference

Without Vision

  • — Claude can't see your UI
  • — You describe screens in words
  • — Manual testing takes hours
  • — Visual regressions go unnoticed

With Vision

  • + Claude sees every pixel, every element
  • + AI clicks, types, scrolls like a user
  • + Tests run in seconds, not hours
  • + Visual changes detected automatically

How It Works

1

AI opens a real browser

Vision launches Chrome with full rendering, just like your users see.

vision.navigate ("https://your-app.com")
2

AI sees what you see

Screenshots are captured and analyzed in real-time.

Page analyzed

Found: 3 buttons, 2 inputs, 1 form

3

AI interacts naturally

Click buttons, fill forms, scroll pages—just like a user.

vision.click("Login")

vision.type("#search", "query")

vision.scroll("down")

4

AI reports what it finds

Results with screenshots, not just pass/fail.

Signup flow works. 3 screenshots captured.

The Tools AI Uses

Available via MCP (Model Context Protocol)

vision.navigate(url)

"Go to the settings page"

vision.click(target)

"Click the submit button"

vision.type(selector, text)

"Fill in the search box"

vision.screenshot()

"Show me what you see"

vision.extract(selector)

"Get all product prices"

vision.wait_for(element)

"Wait for the modal to appear"

What You Can Do

E2E Testing

"Test the checkout flow with a valid credit card"

Form Filling

"Fill out the support ticket with this info"

Data Extraction

"Scrape all product names and prices"

Visual QA

"Does the mobile layout look correct?"

Get Started

One gem. Zero config. Works in 30 seconds.

# Add to your Gemfile

gem 'brainzlab'

# Run bundle

$ bundle install

# That's it. Browser automation is ready.

Coming Soon

We're building Vision right now. Follow along to get early access and help shape the features.

Follow the Journey View All Products