Claude Code for creators: AI editing is here
Brooke Lacey, a 200K+ TikTok creator, demonstrates how she uses Claude Code to automate short-form video editing by leveraging existing computer tools like FFmpeg, Whisper, and Python without uploading files to the cloud. She trains Claude on her TikTok corpus to understand her editing style, facial expressions, and comedic timing, then uses it to generate jump cuts, captions, scripts, and mine comments for new content ideas.
Summary
In this episode of "How I AI," Brooke Lacey shares her innovative workflow for creating high-quality short-form TikTok content using Claude Code. Lacey, an experienced video creator with 25+ years in tech, struggled with the tedious process of manually editing jump cuts to match her comedic timing and facial expressions—a task that would consume hours per video. She discovered that Claude Code could orchestrate existing computer tools (FFmpeg for video processing, Whisper for speech-to-text, and Python) to automate the editing process entirely locally without uploading files to external servers.
The technical approach works by: (1) having Claude detect silence and analyze word timing from Whisper to identify precise cut points, (2) tracking facial features frame-by-frame to adjust crop sizes (wide, mid, close), (3) placing eight strategically-chosen cuts based on a TSV file that serves as the complete edit specification, (4) automatically inserting app screenshots and graphics at relevant moments, and (5) generating one-word-at-a-time captions with proper synchronization. All of this happens in approximately 15 seconds, converting 5 minutes of raw footage into a polished 15-second vertical video. Because the edit exists as a text file rather than a traditional video project, the same command generates horizontal and 60-second versions automatically.
To train Claude on her style, Lacey downloaded her full TikTok corpus to Google Drive, allowing Claude Code to access her previous videos and their comment threads. She taught Claude to recognize her specific comedic hooks, cadence, hand gestures, pauses, and laugh patterns—essentially encoding her personality into the system. This enables Claude to not only improve the editing of new videos but also generate scripts for new content by analyzing what resonated with her audience based on comment data. Lacey emphasizes that this required foundational knowledge of video editing principles; she's not a novice being lifted by AI but rather an experienced creator who structured her domain knowledge in a way Claude could operationalize.
Lacey currently posts approximately once per week, largely because her existing content continues to circulate and gain engagement. She's now exploring live streaming and creating advertisement videos for her Zosha app (a free AI learning platform). When Claude produces unsatisfactory results, she uses a personal operating system she built with Claude that includes memory lanes, status files, and project specifications—allowing her to debug by directing Claude to review prior instructions and global system state.
Key Insights
- Lacey solved the problem of manual jump-cut editing by having Claude use both silence detection and word-timing data from Whisper together, because neither tool alone can place cuts at the precise moment a word ends—requiring sub-word-level accuracy for comedic timing
- The entire video edit can be represented as a text file (a TSV with start/end times, shot sizes, and labels), which enables generating multiple versions (vertical, 30-second, 60-second, landscape) from a single raw take by rerunning the same command
- Lacey had to explicitly teach Claude to recognize her comedic cadence by pointing out specific vocal patterns (such as a particular laugh) and explaining that the hook works by catching attention through tone, not just content
- All processing happens locally on the computer using existing tools (FFmpeg, Whisper, Python)—no files are uploaded anywhere, only instructions are pulled, and the files remain on the user's machine throughout the entire process
- Lacey leverages her 25+ years of video creation experience to orchestrate Claude effectively; she states that while AI can now automate the execution, creators still need foundational knowledge of video editing principles to direct the AI properly
Topics
Transcript
[0:00] No timeline, no project file, no editor. You figured out a way to use Claude code to encode editing that really fits the form, which is short-form content. The file's in the computer, it doesn't upload anywhere. The only thing that's happening is the instructions [music] are being pulled and it's doing all of the things. It's just running the computer. And [music] not only that, but it's listening and goes, "Oh, I just heard you say cut that out." And then it also was like, "You repeated this a couple of times and I took the one that [0:31] was the best cadence." And I was like, "What?" >> And what just happened is Claude descended upon our…
Full transcript available for MurmurCast members
Sign Up to AccessMore from How I AI
I Turned a $50 Photo Frame Into an AI Family Dashboard
A creator transformed a $50 Amazon photo frame running Android into a customized AI-powered family dashboard using OpenAI's Codeex and web technologies. The dashboard displays family schedules, photos, NASA images, news, and individual checklists for each family member, with updates automatically posted via an API connected to a grabbot that monitors calendars and emails.
Muse dodged every trap in this billing ticket
A billing ticket resolution task was used to benchmark agentic AI performance, and Muse excelled by successfully navigating multiple challenges including account linking errors, duplicate refund issues, and prompt injection attempts. The agent demonstrated superior performance by investigating thoroughly, fixing the account link, issuing the correct refund, updating the plan, and marking the ticket as resolved.
Meta's Muse is the best AI agent I've tested
The speaker provides their initial impressions of Meta's Muse AI agent, declaring it the best agentic AI they've tested. Muse excelled across four different knowledge work tasks including billing management, customer responses, inbox triage, and research, outperforming alternatives.
Meta's Muse: best AI agent, worst designer
The speaker tests Meta's AI model on video editing and finds it excels at technical tasks like cutting and timing but significantly underperforms on creative work like design and overlays. They conclude that while the model is strong for coding and agent tasks, it falls short compared to Claude Opus 3.5 for creative applications.
No two Muse agents are the same
The speaker discusses how Muse agent setups are highly personalized and variable, emphasizing the importance of understanding underlying system architecture rather than obsessing over individual details. They describe adopting a more flexible, liberating approach to design that accepts good outputs without requiring perfectionism.