Descript is a video and audio editing platform built around transcription and text-based editing. In 2026, its AI co-editor Underlord has become a central part of the workflow, helping users perform editing tasks through natural-language instructions.

What Descript does

Descript automatically transcribes recordings and lets users edit video by editing the transcript. Deleting or rearranging words changes the corresponding media.

Underlord can perform or assist with tasks such as filler-word removal, clip selection, show notes, scripting, visual edits, and other editing operations.

Key capabilities to know

  • Descript automatically transcribes recordings and lets users edit video by editing the transcript. Deleting or rearranging words changes the corresponding media.
  • Underlord can perform or assist with tasks such as filler-word removal, clip selection, show notes, scripting, visual edits, and other editing operations.
  • Descript's current workflow also includes recording, collaboration, captions, thumbnails, publishing, translation, and content repurposing.
  • Underlord can use different AI models for different jobs, giving users more choice over the reasoning or generation model used for a request.
  • Descript's platform is aimed at creators, podcasters, marketers, educators, and teams that need to turn recordings into polished content efficiently.

How the workflow works

The text-based model changes the starting point for editing. Instead of moving clips around a timeline first, users can begin with the transcript and identify what should stay, what should be removed, and what needs to be improved.

Underlord adds a conversational layer on top of that structure. The user can describe an edit, review the result, and continue iterating while retaining access to conventional editing controls.

  1. Record or import video and audio.
  2. Review the automatically generated transcript and clean up the text.
  3. Use text editing or Underlord instructions to remove unwanted sections and shape the story.
  4. Add captions, layouts, audio cleanup, clips, and platform-specific exports before publishing.

Practical use cases

  • Podcast recording, editing, and publishing.
  • YouTube tutorials and product demonstrations.
  • Short-form social clips from long recordings.
  • Online courses and educational videos.
  • Marketing and customer-story videos created from interviews.

What to consider before adopting it

AI editing can save time, but it does not remove the need for editorial judgment. Automated cuts may be technically correct while still weakening the narrative, context, or personality of the final video.

Teams should also consider brand templates, captions, permissions, music licensing, and review processes when publishing at scale.

Bottom line

Descript's distinctive value is the combination of transcript-based editing with an AI agent that can operate on the underlying media project. That makes it useful for teams that want editing to feel closer to working in a document than in a traditional timeline-first editor.