Here is the Jarvis edit! I told him free tools only!
Here’s what he did:
1. He generated a transcript first, then used FFMPEG to cut dead space, trim the ums & ahs, and drop the sections I flagged after I watched it.
2. After that he scraped images from the Aethir site for overlays, connected to the Remotion API (
remotion.dev/) for animations and graphics, and I let him inject some personality.
This took 3-4 iterations. I felt myself spending more time on this than I was expecting and had to pull back or the test wouldn’t have been fair.
Decent result and shows the capability right now, but there’s a still a lot of ground to cover for this to be a one-shot solution.
Also, I could have saved myself a lot of time by preparing a stronger plan and prompt for Jarvis before setting him off on the video editing job. Will be interesting to see how much better the next video will be when I incorporate these prompt learnings.
Good job, Jarvis