Building an agentic video editing prototype in Rust using GPUI and wgpu

Name: Building an agentic video editing prototype in Rust using GPUI and wgpu
Availability: InStock
Author: LOVELYZOMBIEYHO

by LOVELYZOMBIEYHO·Apr 19, 2026·1 point·0 comments

Visit Project View on HN

AI Analysis

●●SolidWizardryBold BetShip It

Rust video editor with agentic silence cutting, but Descript does this already.

Strengths

•GPUI and wgpu stack enables native GPU-accelerated performance in Rust.
•Agent Client Protocol integration allows conversational timeline manipulation directly inside.
•Local Whisper models enable offline subtitle generation and translation workflows.

Weaknesses

•Alpha status with no prebuilt binaries limits immediate adoption significantly.
•Core silence-cutting features already exist in mature tools like Descript.

Post Description

Hi HN,

I’ve been building a video editing (NLE) prototype in Rust, exploring a different workflow: prompt-assisted editing.

Instead of scrubbing through the timeline to find silence and manually cutting clips, you can just type:

“Help me to cut silent parts”

“Help me to cut silent parts below -14 dB”

and the editor analyzes the timeline and removes them automatically.

I mostly work with interview-style and knowledge-based videos, where a lot of time is spent trimming silence and cleaning up rough cuts. The goal is to see whether this kind of workflow can significantly speed that up.

I’m also experimenting with:

- "Removing repeated subtitles" - "Cutting gaps where no one is speaking" - "Suggesting B-roll based on subtitles (using an LLM)"

The project is built in Rust, using GPUI for the UI, wgpu for rendering, and gstreamer + ffmpeg for preview and export. I’m still exploring the architecture and performance tradeoffs, especially around timeline processing and NLE-style editing operations.

Curious if anyone here has worked on NLEs or media tooling in Rust, or has thoughts on using Rust for this kind of workload.