Cutroom

A project starts with an upload. The web app accepts common video and audio formats, including MP4, MOV, WebM, MKV, M4V, MP3, WAV, M4A, AAC and FLAC. Before any editing happens, the agent probes the video and audio streams, transcribes speech with word timestamps, and builds a visual index of the footage using contact sheets that span the source, so shot changes are visible alongside the transcript.

Next you direct the edit in conversation, the way you would brief a human editor. You might ask it to cut the slow opening and start on the result, or to turn a long interview into a vertical short with a faster hook. The agent translates that intent into exact time ranges, caption rules, grade settings and title overlays, and reports back with what it found - for example, a stronger hook at a specific timestamp - along with a plan ready for review.

You then inspect the plan and approve a preview render. The worker verifies the timeline and media before creating the final master, which requires a separate approval. Aspect ratio, pacing, captions and export rules all live in the plan, so you can revise any of them in the chat and generate a new version rather than starting over. Paid plans include final master download and commercial use.

• •Word-aligned cuts - Dialogue edits are placed on word timestamps with short safety handles around each cut, so filler and silence removal does not clip syllables. Brief audio fades at edit boundaries avoid clicks.
• •Reviewable edit plan - Every request becomes a deterministic plan you can read before anything renders: the number of cuts, caption language, overlays and grade. Preview and final export have separate approval gates, so an expensive render never runs on a plan you have not seen.
• •Full finishing pipeline in one agent - The same conversation handles transcription, filler and silence cuts, burned-in captions, color and audio polish, and motion overlays such as title animations.
• •Multi-format output - Projects can keep the source aspect ratio or target landscape 16:9, vertical 9:16 and square 1:1. Vertical edits can be reframed with a front-loaded hook and bold readable captions for social platforms.
• •Deterministic rendering - Renders are built from FFmpeg-based edit decision lists with consistent timebases, short audio fades and captions applied last, producing outputs the worker can verify. Rendering runs in isolated project workspaces with signed worker jobs, path isolation and size limits.
• •Revision history - Source media, plans, previews and final masters remain attached to the project, so you can iterate safely and return to an earlier version instead of re-uploading and re-directing the edit from scratch.

Cutroom suits creators and marketers who publish regularly but do not want to learn a professional timeline editor. Founders and solo creators recording talking-head updates can strip filler words and pauses while keeping natural breathing room, then add captions in the same pass. Social media teams can turn longer recordings into 9:16 shorts with a faster hook and readable burned-in captions.

Product marketers will find the product story workflow useful for launch videos, where claims need to be matched to product footage and title overlays need consistent brand timing. Educators and SaaS teams producing screen tutorials can trim waiting time, emphasize key moments and make each instruction easier to follow.

It also fits people who want AI speed but distrust fully automatic editors, since the plan is visible and each render needs approval. Project browsing and review work on mobile, though a larger screen is recommended for detailed timeline inspection. The Lite plan is aimed at trying the complete workflow, Pro at weekly publishing, and Ultra at production teams with higher monthly volume.