Video Use: Edit Videos by Chatting with AI

Ever wished you could edit a video just by telling someone what you want? Like, "cut out the silence" or "add a zoom here"? Well, now you can – sort of.

Video Use is a new open-source tool that connects a web-based video editor with a large language model (LLM). The idea is simple: you chat with an AI assistant, and it edits the video for you. No timeline dragging, no keyframes. Just describe what you want.

What It Does

Video Use is essentially a browser-based AI agent that controls a video editing interface. You give it natural language commands (like "trim the last 3 seconds" or "add a fade-in"), and it breaks those instructions into editing actions. The AI sees the video editor's UI, clicks buttons, sets parameters, and runs the edits – all autonomously.

Under the hood, it's powered by the same kind of LLM reasoning that drives tools like Browser Use (which is also by the same maintainer). The repo is a fork or extension of that idea, but focused on video editing workflows.

Why It's Cool

No API integration needed. Video Use isn't calling a remote video processing API. It works by literally controlling a web application in a browser. That means it works with existing video editors, not just custom ones. The demo shows it running on a popular web-based editor (likely Kapwing or a similar tool), which means the approach could generalize.

Natural language as a UI. Instead of memorizing keyboard shortcuts or digging through menus, you just say what you want. For devs, this is like building a "voice interface" for video, but in text. It's a neat proof of concept for how LLMs can make complex creative tools more accessible.

Open source and hackable. The code is straightforward. You can see how the AI decides what to click and when. If you want to extend it to other editors or add custom commands, you can.

Real use cases. Think about these:

• Automating repetitive cuts in interview footage (remove silences, trim intros).
• Creating quick social media clips from longer videos.
• Non-technical users (like content creators) who don't want to learn editing tools.

How to Try It

1. Clone the repo:

   git clone https://github.com/browser-use/video-use.git
   cd video-use
   

2. Install dependencies (Python + Playwright):

   pip install -r requirements.txt
   playwright install
   

3. Set up your LLM API key (OpenAI or Anthropic, check the .env.example).

4. Run the agent with a command like:

   python run.py --task "Add a zoom-in effect at 0:10"
   

The README has more details and a demo video. You'll need a web-based video editor open in a browser – the AI will take it from there.

Final Thoughts

Is Video Use production-ready yet? Probably not for complex projects. But it's a clever demo of where AI and creative tools are heading. For devs, it's a fun weekend project to hack on – try adding support for your favorite editor, or tweak the prompt to handle more advanced edits.

The fact that it works by controlling a browser (instead of needing a custom API) makes it surprisingly flexible. If you've ever wanted to automate video editing without learning scripting, this is a glimpse of that future.

Let us know what you build with it.

Follow us on X: @githubprojects