Hi all,
I’m experimenting with using LangChain to automate parts of a short-form content pipeline, where the actual editing is still done in capcut.
Idea
I’m trying a lightweight agent-based setup:
-
One agent collects and organizes raw video clips
-
Another generates titles, captions, and hashtags
-
A script-based step (FFmpeg) preps files for editing
-
Final editing happens in free.
-
After export, another agent handles renaming and uploading
Questions
-
Does this kind of workflow make sense with LangChain, or is it unnecessary complexity?
-
Would LangGraph be a better fit for managing these steps?
-
Has anyone built similar human-in-the-loop media pipelines?
Just trying to reduce repetitive work and keep capcut focused only on creative editing. Would appreciate any feedback or suggestions.