
The fastest way for an agent to use a browser is to write code.
The Hook
“(No spoken dialogue — ambient typing sounds and subtle UI clicks)”
Problem Agitation
“(No spoken dialogue — ambient typing sounds and subtle UI clicks)”
Product Reveal
“(No spoken dialogue — ambient typing sounds and subtle UI clicks)”
Feature Teaser
“(No spoken dialogue — ambient typing sounds and subtle UI clicks)”
Synthesizable in Remotion · key components of the ai-dev-browser-automation template.
A persistent split-screen layout with a dark-themed terminal on the left and a light-themed browser on the right, maintaining a clear visual hierarchy.
Utilize Remotion's <AbsoluteFill> and <Sequence> components to create two distinct, fixed-width columns. Implement a custom `Terminal` component for the left side, handling text typing animations and scrollable content. The right side will host a `BrowserFrame` component, capable of rendering web content and overlaying UI elements.
Text appears character by character in the terminal, simulating real-time typing, often followed by a rapid display of command output.
Employ Remotion's `interpolate` function with `spring` for character-by-character text reveal. Use a `delay` prop for each line or block of text to simulate processing time. For command output, a faster `spring` or `linear` interpolation can be used to quickly reveal the results.
The browser window dynamically responds to commands from the terminal, including navigation, modal pop-ups, and automated data entry into web forms or spreadsheets.
Create a `BrowserWindow` component that can receive props to simulate URL changes, display modal overlays, and render dynamic content. For data entry, use `interpolate` to animate the appearance of text within spreadsheet cells, mimicking rapid typing or pasting. SVG filters can be used for subtle hover states or focus indicators on UI elements.
Purpose. Introduce the AI agent and the user's initial request, establishing the core problem of manual browser tasks.
Execution. The split-screen terminal and browser appear, with the user typing a natural language command into the terminal.
Purpose. Demonstrate the AI's ability to connect to and control the browser, highlighting the 'dev-browser' tool.
Execution. Terminal output confirms skill loading and browser connection, followed by a permission modal in the browser that is automatically accepted.
Purpose. Showcase the AI performing complex browser actions, specifically creating and populating a Google Sheet.
Execution. The browser navigates to Google Sheets, creates a new spreadsheet, and then rapidly populates it with data, all driven by terminal commands.
Purpose. Conclude with the successful completion of the task, emphasizing the efficiency and power of the AI agent.
Execution. The completed, formatted Google Sheet is displayed, and the terminal confirms all actions are finished, including a final screenshot.
Objective
An AI-powered code refactoring tool that optimizes legacy codebases.How the recipe adapts
The split-screen layout will feature a terminal on the left where a user types a command like 'refactor-legacy-code --project=my-old-app'. The right side, a web-based IDE, will show a block of spaghetti code. As the terminal processes, the IDE will visually animate the code transforming into a clean, optimized version, with new lines appearing and old lines fading out, demonstrating the refactoring in action. The same typing and rapid UI update primitives will be used.Scene durations (4 scenes, 30.4s)
Key Remotion components
Agentic CLI for coding, building apps, and automating workflows.

Sakana AI launches Fugu, a multi-agent orchestration system served through a single model API, led by its Fugu Ultra model.

Get paid to wait The Claude Code spinner might be the most watched line on Earth. So I turned it into an ad marketplace. Advertisers bid on it. You keep 50% of the money. Install the extension → ge…