I Tried Fully Automating My Video Editing With AI (RAW RESULTS)
Published
Should I watch this?
Skip to the real content 1:23Your full summary is ready below. Read on while we finish your watch recommendation.
Your summary is ready. The watch recommendation is next.
Worth watchingYes
The experiment usefully shows where automated editing fails and how a clean manual cut, storyboard approval, and iterative Hyperframes or Remotion graphics produce a workable AI-assisted launch video—not one-click automation.
The main answer
Yes—AI can help produce a polished launch video, particularly its motion graphics, but this experiment does not demonstrate one-click editing or replacement of a skilled editor. The successful workflow was to write a short script, record it, manually finish a clean talking-head cut, then use Claude with Hyperframes or Remotion to create and refine animated graphics. It took roughly six hours and, by the creator’s estimate, about 20 rounds of prompting. His “97% as good” conclusion is subjective, not a measured result; he still acknowledges that his human editor’s work is better.
Key takeaways
- Separate editorial work from graphics production: selecting takes and building a coherent story are different tasks from animating product screenshots.
- The initial attempt to extract a polished short video from several minutes of unscripted footage produced awkward cuts and technical problems.
- For predictable short-form results, start with a tight script and clean recording rather than expecting AI to discover the entire story.
- Approve storyboards and contact sheets before rendering. Fixing the plan is easier than repeatedly rebuilding finished videos.
- Specific instructions and relevant production vocabulary helped the creator communicate what he wanted.
- Hyperframes performed best initially; Remotion improved substantially after refinement. These are results from one novice’s experiment, not a controlled comparison of the tools.
- Code-based graphics made broad styling changes easier, although updating the branding still required rerendering.
- The closing claim of no extra cost beyond Claude overlooks the roughly $300 Resolve purchase shown earlier. Existing subscriptions, hardware and working time also complicate the cost claim.
Summary
The creator spends a day testing AI-assisted video production against a launch video made by his team’s editor. He initially tries Claude-controlled DaVinci Resolve, Hyperframes and Remotion in parallel. Early results include poor dialogue cuts, buggy graphics and uneven styling. Progress improves once he narrows the task: he records a purpose-written script, manually completes the core edit, and asks AI to produce animated overlays. After repeated revisions, audio additions and a branding update, he considers both code-based versions impressive. The practical conclusion is human direction plus AI assistance—not autonomous editing.
What’s being promoted?
Explicit paid sponsorship: Whisper Flow sponsors the video. The creator demonstrates its dictation, formatting and correction features, contrasts it with built-in phone dictation, and offers a free month of Pro through his description link. He explicitly asks viewers to name him as their referral because that could encourage future sponsorships. This establishes a commercial incentive, but the transcript does not establish whether he receives commissions or whether the comparison represents typical performance. Self-promotion: Creator Grid, which he identifies as his own business, is the subject of the entire experiment. Viewers repeatedly hear its sales pitch: managed creator campaigns, no retainer or subscription, payment for approved content or views, and an invitation to get in touch. This makes the tutorial a substantial product-marketing vehicle as well as an educational experiment. Claims about customers, savings and more than 10,000 creators are not independently verified here. He also promotes FireCut indirectly by identifying it as a tool his team built, although he admits its first cut still needed manual correction. Other products: Claude, Grok, Hyperframes, Remotion, DaVinci Resolve, Canva, Epidemic Sound and Apple hardware receive exposure. The transcript does not establish sponsorship or affiliate relationships for these mentions. Their presence alone is not evidence of undisclosed payment. Broader agenda: The video also builds the creator’s reputation as an approachable AI experimenter and directs viewers to another video. Dramatic lines such as “video editors are cooked” create excitement but overstate the demonstrated outcome. However, showing failures, acknowledging repeated prompting and praising the human editor provide meaningful counterbalance. Commercial motives are clear; deliberate deception is not established. The transcript alone cannot verify the visual quality or fairness of the final comparison.
