AI video editing
I still record in Riverside and Descript. Most of my editing now happens in a terminal, with Claude or Codex driving HyperFrames, Tesseract, and FFmpeg.
These are the edits I wrote up with my real prompts, the brand rules I gave the agent, and what went wrong along the way. One took zero revision notes. One took three rounds of corrections. One I ran from my phone.
Generated media is the seasoning, not the meal. I use AI to dress up what I shot, not to replace me on camera.
New to editing with an agent? Start with how my workflow fits together, then the one-shot edit, then the one I ran from my phone.
My workflow · October 10, 2026 · 10 min read
Ryan Doser and I talked AI video editing on AI Rabbit Holes. I still record in Riverside and Descript, but Claude, HyperFrames, and Tesseract do most of my editing now.
Claude + Tesseract · October 4, 2026 · 9 min read
I sent one prompt and gave zero revision notes. Here are the real prompts, Linear ticket, brand rules, and outputs behind a 2:11 ChatGPT dot vs Grok Bot CarPlay video, and how to set up your own version.
Codex + HyperFrames · July 25, 2026 · 12 min read
The real prompts, screenshots, corrections, and tools behind a 31.7-second vertical video I edited with Codex and HyperFrames from my phone.
Newest first.
My workflow · October 10, 2026 · 10 min read
Ryan Doser and I talked AI video editing on AI Rabbit Holes. I still record in Riverside and Descript, but Claude, HyperFrames, and Tesseract do most of my editing now.
Claude Code + HeyGen · October 9, 2026 · 9 min read
I filmed it in 33 minutes and spent the evening editing it with Claude Code, HeyGen, and FFmpeg. Here's every prompt I sent and what I'd do differently next time.
Claude + Tesseract · October 4, 2026 · 9 min read
I sent one prompt and gave zero revision notes. Here are the real prompts, Linear ticket, brand rules, and outputs behind a 2:11 ChatGPT dot vs Grok Bot CarPlay video, and how to set up your own version.
Tesseract · September 21, 2026 · 10 min read + videos
My exact prompts, three before-and-after edits, and the caption, sound, and camera rules I saved for the next Tesseract video.
Codex + HyperFrames · July 25, 2026 · 12 min read
The real prompts, screenshots, corrections, and tools behind a 31.7-second vertical video I edited with Codex and HyperFrames from my phone.
Not for most of my edits. I point Claude or Codex at the raw file and have it drive free tools like HyperFrames, Tesseract, or FFmpeg in a terminal. I still record in Riverside or Descript, because they give separate tracks and good audio when two people record together.
A powerful one with local access. Rendering uses a lot of memory even in the terminal, and some of these tools need a Mac. A MacBook Air or a Chromebook will struggle, and a cloud bot like Grok Bot cannot run the tools that require a Mac.
Give it a brand guide for video. Mine is mostly caption rules: where they sit, the fonts, how words animate in, and never covering my face. I keep it in a skill, use it on every edit, and make it a little better each time.
No. Generated media works best as B-roll, a graphic that shows what I am talking about while it is still me on camera. Platforms are cutting back on how fully generated videos earn money, and they all start to look the same.
My notes on AI in schools from 17 years in classrooms: policy, classroom habits, agentic AI, teacher tools, and letters to teachers and principals.
Every AI Rabbit Holes episode I've joined Ryan Doser on, written up in my own words: daily driver models, open source AI, video editing, and Grok Bot.
Everything I have written about Grok Bot: my review, setup, usage limits, email, and routines.
@theaaron
I'm @theaaron on X, where I post the edits and what I'm testing as I go.