Stephen A. Hart
I’m a heavy Descript user with 40+ custom saved layouts for a single client and over 50 fully edited videos in that series (each about 5–7 minutes long). Every video follows a consistent visual language: layouts, lower thirds, zooms, transitions, and pacing that I’ve already “taught” Descript through my past work.
It would be game changing if Underlord (or a similar Agentic AI feature) could:
-> Analyze my existing edited library
-> Study my 50+ completed videos plus the 40+ saved layouts, scenes, and styling choices.
-> Learn my recurring patterns: how I time jump cuts, how I use zooms/emphasis, and where I insert titles, overlays, etc.
Auto-edit new videos “in my style”
Take a new 5–7 minute recording and generate a first-pass edit that:
  • Applies the right layouts from my saved set based on content and context.
  • Proposes additional layout variations that match my visual style.
  • Inserts titles, lower thirds, and transitions in the places I usually add them.
  • Act like a layout-aware editing assistant, not just a clip trimmer
Let me choose a “Style Profile” (e.g., Client A – YouTube Series) so Underlord knows which project/library to learn from and apply.
Give me a preview timeline with clearly marked “AI decisions” so I can tweak or approve them AND over time, allow Underlord to refine its choices based on what I keep, move, or delete.
This would turn Descript from an AI that helps with pieces of the edit into a true agentic co-editor that understands my visual system and applies it at scale.
For creators and agencies producing dozens or hundreds of on-brand videos for the same client, this kind of layout-aware, style-learned automation would massively reduce edit time while keeping every video consistent with the brand’s look and feel.