Descript's new AI editor does the grunt work

Editing video is rarely about creative genius. Most of the time, it is about agonizing over filler words, manually centering webcam frames, and scrubbing through timelines to cut bad takes. Descript wants to end that grind with an agentic approach.

The company has introduced Underlord, a conversational AI co-editor designed for video and podcast creators. Descript pitches the experience as vibe editing: instead of clicking through endless menus or searching YouTube for tutorials, you direct an assistant that already knows the software inside out.

Delegating the tedious grunt work

Underlord is built to watch your raw video, read the transcript, and actively suggest edits. Creators can pitch an idea, ask for a script draft, and refine it through natural language feedback.

The assistant handles heavy multi-step lifting in a single prompt. It can center speakers automatically, bleep out profanities, generate lower thirds, and repurpose long webinars into promotional clips for social feeds. It also turns static presentations, blog articles, and screen recordings into watchable video layouts.

Core Descript capabilities remain anchored to text-based editing, meaning deleting text still deletes video. But Underlord now ties in automated filler-word removal, gaze adjustment, and studio-grade audio enhancement under unified conversational controls.

Pricing starts with a free tier offering one media hour per month, 100 one-time AI credits, and watermarked 720p export. Paid tiers run at $16 and $24 per month, unlocking watermark-free 1080p or 4K exports, up to 30 media hours, voice cloning, and extended AI credits.

Why it matters

For solo creators, podcasters, and small marketing teams, Underlord solves the single biggest bottleneck in content production: editing fatigue. If your workflow revolves around talking heads, interviews, or software walkthroughs, this is an immediate productivity boost that requires almost zero traditional editing skills.

Across the creative software landscape, Descript is redefining what a modern video editor looks like. Instead of getting bogged down in generative video parlor tricks, they are applying agentic AI directly to workflow automation. Competitors clinging to classic timeline interfaces will face mounting pressure as users realize they prefer chatting with an assistant over keyframing.

We think Descript played this smartly by leaning into its transcription-first strength. The risk, however, lies in creative nuance. When an AI co-editor misinterprets conversational pacing or humor, directing it back on track through chat prompts can sometimes feel more frustrating than making manual cuts on a timeline.