YouTube and Social Media

How to set up AI video editing with Claude from script to published Reel

[Claude] AI Video Editing: The System Behind 2.3M Views 🤖

Video: [Claude] AI Video Editing: The System Behind 2.3M Views 🤖

· based on the channel Andrew Kroeze

AI video editing means letting a model like Claude handle the repetitive parts of post-production — cutting dead air, assembling a rough timeline, writing the title and thumbnail copy, and scheduling the finished file — while you keep control of the creative decisions. A practical version of this workflow runs on four tools: Screen Studio for capture, Claude for editing and packaging, Auphonic for audio, and Zernio for publishing and comment-to-DM automation. Andrew Kroeze demonstrates the full chain in his breakdown, and you can join the live build sessions where these systems are assembled with him by grabbing a seat at the Tuesday AI Build Room.

The result is not a fully autonomous editor. It is a system where a single operator produces YouTube videos, Instagram Reels and static posts without a production team, and where the model does the work that used to require an editor, a copywriter and a community manager.

What an AI video editing stack actually looks like

A working stack separates capture, editing, packaging and distribution so each stage can be automated independently. In the workflow shown, Screen Studio records the screen with automatic zoom and cursor smoothing, which produces footage that already looks edited. Claude then takes the raw file and the transcript, writes the cut list, drafts the title, description and thumbnail text, and hands the package to Zernio, which publishes to YouTube and Instagram on a schedule.

The important shift is that the model is not asked to be creative from nothing. It is given a transcript, a brand kit and a set of instructions, and it returns structured output: which segments to keep, what the hook should say, which caption belongs on which platform. That structure is what makes the output repeatable across dozens of videos instead of working once.

The pipeline behind a 2.3 million view video

Dr. Emeka spent roughly 350 hours refining an AI editing pipeline, and one video produced through it reached 2.3 million views while his team shrank from 8 employees to 1. The pipeline has four stages, and each one removes a specific manual task.

  1. Inspiration research with Apify. Scrapers pull high-performing videos in the niche so the format, hooks and pacing are based on evidence rather than guesswork.
  2. A voice bank. Recorded samples of the creator's voice are stored so generated or re-recorded lines match the original tone instead of sounding synthetic.
  3. The Script Lab interview. Instead of writing a script from a blank page, the creator is interviewed about the topic and the answers become the raw material for the script.
  4. A voice-follow teleprompter. The prompter advances as the creator speaks, so delivery stays natural and the take needs fewer retakes.

Each stage feeds the next, which is why the system survives contact with a real publishing schedule.

Audio cleanup and comment-to-DM automation

Clean audio does more for perceived production value than any visual effect, and two tools cover most of the need. Auphonic normalizes levels, removes noise and balances multiple speakers automatically, while Adobe Podcast Enhance rescues recordings made in untreated rooms. Running raw footage through either one before editing saves the model from trying to cut around audio problems.

On the distribution side, Zernio handles comment-to-DM automations: when someone comments a specific keyword on a post, the tool sends them a direct message with the link or resource. For coaches and consultants this converts engagement into conversations without anyone monitoring the inbox, and it closes the loop between a published Reel and an actual lead.

How the Reel Room turns one video into many posts

The Reel Room is a production structure built around one body and many entry points. A single recorded body is paired with multiple hooks and multiple calls to action, so the same core content ships as several distinct Reels rather than one. Client notes are fed back to Claude, which re-edits the existing footage instead of starting over.

This matters because most creators lose time re-recording instead of re-framing. If a hook underperforms, the fix is a new opening thirty seconds long, not a new video. Trial Reels through Buzzfy let you test variations on non-followers before committing a concept to the main feed, and static posts fill the gaps between video releases using the same brand kit.

The 5 C's for prompting an AI editor

Getting usable output from Claude depends less on the model and more on how the request is framed. The 5 C's cover the essentials: be clear about the task, give context about the audience and platform, constrain the output format and length, provide the raw material such as transcript and brand kit, and confirm the result against a concrete example of what good looks like. Vague requests produce generic edits; constrained requests produce something you can publish with minor tweaks.

Common questions about AI editing systems

Most skepticism about AI video editing comes down to two worries: that the output will sound like everyone else's, and that setup takes longer than editing by hand. The honest answer is that the first few videos take longer, because you are building the brand kit, the voice bank and the prompt structure. After that, the marginal cost of a video drops sharply.

The second recurring question is about team size. Cutting from 8 employees to 1 was not a matter of replacing people with a model overnight; it came from fixing the biggest bottleneck first, which in that case was editing throughput. Automating a stage that was not the constraint produces no visible gain, which is why the advice is to measure where work actually stalls before buying another tool.

Summary

AI video editing works when it is treated as a pipeline rather than a single prompt: capture in Screen Studio, let Claude cut and package, clean audio with Auphonic, publish and automate replies through Zernio. The 2.3 million view result and the move from 8 employees to 1 came from fixing the editing bottleneck first and building reusable assets such as a voice bank and brand kit. Andrew Kroeze, the author of this video breakdown, walks through the whole chain and runs free Tuesday sessions where the systems are built live — reserve a spot at the AI Build Room and bring your own footage.

Key takeaways

  • Andrew Kroeze's workflow records in Screen Studio and lets Claude edit, package and post the video through Zernio.
  • Dr. Emeka spent 350 hours building an AI editing pipeline; one video from it reached 2.3 million views and his team went from 8 employees to 1.
  • The pipeline uses Apify for inspiration research, a voice bank, the Script Lab interview and a voice-follow teleprompter.
  • Auphonic and Adobe Podcast Enhance handle audio cleanup; Zernio automates comment-to-DM replies.
  • The Reel Room produces one body with multiple hooks and CTAs, and Claude re-edits from client notes.

Source: [Claude] AI Video Editing: The System Behind 2.3M Views 🤖 · Markdown version

Questions & answers

Do I need to know how to code to run this AI editing setup?

No. The tools are configured through prompts and settings rather than code, and Claude Code is used to wire steps together, not to write a custom application. The learning curve is in structuring your brand kit and prompts, not in programming.

Will AI-edited videos sound generic compared with my usual content?

Only if you skip the inputs that make the output yours. A voice bank, a written brand kit and examples of your best-performing videos give the model enough material to match your tone, and the Script Lab interview keeps the language in your own words.

How long does it take before the system pays off?

The first few videos are slower than editing manually because you are building reusable assets. Once the brand kit, voice bank and prompt structure exist, each additional video costs far less time, which is where the leverage comes from.

Should I automate editing first or something else in my business?

Fix your biggest bottleneck first. If editing is not the stage where work piles up, automating it produces no visible gain; measure where output actually stalls before adding tools.

Read next