How I Built a Cross‑Platform Audio/Video App in Hours with AI‑Powered CodeBuddy

Background

While working on the TransDuck audio‑video processing SaaS, the author received user feedback about long upload times for large files, high bandwidth costs, and a desire for a dedicated desktop client that could work offline and avoid browser sandbox limitations.

Having never built a desktop client before, the author was torn between PyQt, Electron, and Tauri, and decided to let AI help by using the Tencent Cloud CodeBuddy extension.

Preparation

Install VS Code from https://code.visualstudio.com/.

Search for and install the CodeBuddy extension in VS Code.

Create a new project folder and open it in VS Code.

Requirement Writing

Create a markdown file 0.1.0.md in a docs directory and describe the desired features, technology stack, and UI design.

Implement a cross‑platform audio/video processing desktop app: TransDuck

# Tech Stack
- Node.js 22.15.0
- Tauri 2.0 with Vue frontend
- ffmpeg

# Project Structure
- docs: requirement documents
- example: test audio/video files
- transduck-desktop: source code

# UI Layout
- Left menu, right operation area
- Top task configuration, bottom progress/result display
- Internationalization (Chinese & English)
- Theme: Vue Admin Bootstrap

# Functional Modules
1. Audio format conversion (ffmpeg‑supported formats, same‑folder output, progress bar)
2. Video format conversion (same requirements as audio)

# Other Requirements
1. Project initialization
2. ffmpeg integration (bundle with the app)

Configuring CodeBuddy

Enable Auto‑run and Auto‑modify files so that CodeBuddy can execute commands and edit files directly after the user submits a request.

Development Process

Paste the requirement markdown into CodeBuddy’s input box and select the default model.

CodeBuddy scans the project, creates the directory structure, scaffolds a Tauri + Vue project, and installs dependencies.

When errors appear (e.g., missing commands, build failures), CodeBuddy automatically proposes fixes and applies them.

The developer can approve or reject each step; otherwise the process continues autonomously.

The AI‑assisted workflow repeatedly performs: task planning → code generation → execution → error fixing → next task.

Key Milestones

Project initialization and basic UI layout completed.

Functional testing revealed issues such as file‑selection not working, unusable output format fields, conversion button not triggering, and missing file‑open actions.

CodeBuddy was used to fix these problems, add icons, improve component styles, add an "About" page, refine internationalization UI, and polish progress‑bar and result displays.

Important Tips

Do not keep a single CodeBuddy conversation for too many tasks; start a new session when the current one becomes overloaded to avoid accidental code corruption.

Additional Capabilities

CodeBuddy can also generate commit messages automatically and, via the MCP marketplace, install external services to enhance the development workflow.

Conclusion

The entire challenge took about four hours, whereas a manual Tauri learning and development cycle would have taken several days. The author not only delivered a functional cross‑platform desktop client but also learned Tauri fundamentals through AI‑guided commands and error resolution. The project will be open‑sourced and continuously maintained.