🔍 Read the full analysis: From AI Idea To Production: Start With A Simple First Video on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Thorsten Meyer says a hand-drawn video concept became a live style in hours, but its first full video was slow and visually sparse. After measuring frame coverage, stillness and camera movement, he added checks to the production process; the account is a personal report, not an independent evaluation.
Thorsten Meyer says he took a hand-drawn video idea from an evening prototype to a live version within hours, then revised the production process after the first full video looked slow and sparse. In a November 6 account on ThorstenMeyerAI.com, he describes adding human approval, independent AI code review and measurable checks so that future videos can be stopped before rendering when they miss basic visual standards.
Meyer says the idea came from seeing a whiteboard-style explainer he liked. To avoid copying it, he says he recorded only broad style traits: hand-drawn lines, a continuous visual world and large numbers used as color accents. He says the team used no drawings, assets, layouts or code from the reference video. Claude built the drawing engine, using colors from Meyer’s design system and an openly licensed handwriting font.
A separate system, OpenAI’s Codex, reviewed the code rather than building it. Meyer says Codex identified five issues in its first review, and that each was fixed with a test designed to fail on the old code. He also describes watching a roughly one-minute test video and approving it before publication. That version appeared in an article about local votes against data centers, according to his account.
The first full video in the new style, produced the next morning, did not meet Meyer’s expectations. He says it had too much empty space, icons that did not support the narration and images that stopped moving once drawn. Rather than leave the assessment at personal taste, he measured frame coverage, stillness and camera movement. Across the first three scenes of the rebuilt video, he reports that filled-frame coverage rose from 15% to 66%, stillness fell from 29 seconds to zero, and camera moves increased from three to 27. These figures are his measurements for those scenes, not results from an independent review.
Infographic 01 · The new video look
One evening, one boring first video, one measured fix
Speed was not the point. The steps in between were: a second AI that reviews, a human who says yes, and numbers that became gates.
Share of the frame filled with content on the first three scenes after the fix. Before: 15%.
From an evening idea to a measured fix
Relative times, this week
- EveningA video I likedClean room: only a list of style traits, no drawings, assets or layouts
- Same eveningOwn engine, first reviewClaude builds; OpenAI’s Codex reviews; 5 findings fixed, each with a test that fails on the old code
- Same eveningA human says yesOne-minute test video approved and published
- Next morningFirst full video: slow and boringToo much white space, icons that did not fit, a picture that stood still
- Hours laterMeasured fix, then gatesThe first video in the improved look goes live
Source: my own build notes for the new video look, this week (engine history, review round, approval, critique, before/after measurement of the first three scenes).
Powered by Thorsten Meyer AI
Turning Visual Critique Into Checks
The account shows how an AI-assisted creative prototype can be treated as a production process rather than a one-off demonstration. Meyer’s approach combines human judgment with automated review: a person approves the output, while code tests check for specific defects before a video is rendered. The distinction matters because a system that runs successfully can still produce an unsatisfactory result.
Turning subjective feedback into defined checks may help catch recurring problems earlier. Meyer says the new gates stop scenes that are too empty and flag objects that do not illustrate what is being said. Those checks do not establish that a video is engaging, accurate or legally safe; they address a limited set of criteria. The figures therefore describe process changes, not proof of overall quality.
Meyer also uses other examples to argue that passing a test or completing a task is not enough. In a separate weight-edit experiment, he says a model completed 12 of 15 legal benchmark answers, compared with four before the edit, while its legal score remained zero. He also reports a code review in which all 12 of the author’s tests passed but four of a reviewer’s six checks failed. These are examples from his own work, and the account does not provide enough detail to independently assess the experiments.
Products Worth Considering
Samsung Galaxy Tab A11 Wi-Fi Tablet, 8.7-Inch, 64GB, 4GB RAM, Gray (SM-X133, International Version)
Samsung Galaxy Tab A11+ 11-Inch Tablet, 6GB RAM, 128GB Storage, Expandable, Gray
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
As an affiliate, we earn on qualifying purchases.
From Prototype to Live Video
The timeline in Meyer’s account is short: he saw a video style he liked, built a distinct drawing engine that evening, approved a test clip, and published that version. The following morning’s full video exposed shortcomings that had not been apparent in the brief demonstration. The improved look was then used in a video about the Nasdaq’s worst day in 19 months, as described by Meyer.
The article frames this sequence as a personal method for moving AI ideas into production. Its safeguards include separating creation from review, writing tests that demonstrate a defect existed before a fix, requiring a responsible person to approve output, and measuring specific properties instead of repeatedly adjusting by feel. Meyer also recalls an August monitoring failure: a guard reportedly generated about 1,300 warnings in a log that nobody read, while an alert email did not go out. The example underscores that monitoring has limited value if warnings do not reach a person.
“A demo that runs is not a decision.”
— Thorsten Meyer
What the Report Does Not Establish
The account is Meyer’s own description of his tools, measurements and decisions. It does not include an independent assessment of the videos, the code review, the quality gates or the legal benchmark. The reported visual measurements cover the first three scenes of the rebuilt video; the account does not give a broader sample or a comparison across multiple productions.
It is also unclear how the frame-coverage and stillness thresholds are set, how often the checks produce false alarms, or whether the gates can assess whether a visual accurately represents narration. Meyer says objects that illustrate nothing spoken should trigger a warning, but he does not detail how that relationship is evaluated. The legal-model example is separate from the video workflow, and the available account does not provide the benchmark’s full methodology.
How the Production Rules Will Be Used
Meyer says the visual measurements have become production gates: scenes judged too empty can be stopped before rendering, and unsupported visual objects can trigger a build warning. The article does not set out a schedule for further releases or describe planned changes to the checks.
The next practical test will be whether the process works consistently on later videos, not only on the examples Meyer describes. Readers can compare the published video with the reported workflow, while broader conclusions about reliability, quality and the effectiveness of AI review remain dependent on more evidence.
Key Questions
What happened in the reported project?
Thorsten Meyer says he built a hand-drawn video engine, approved a short test clip and published it, then revised the process after the first full video looked visually sparse and slow.
How did the video change after revision?
For the first three scenes of the rebuilt video, Meyer reports frame coverage rising from 15% to 66%, stillness falling from 29 seconds to zero, and camera moves increasing from three to 27.
What checks were added?
Meyer says scenes that are too empty can be stopped before rendering, while objects that do not illustrate the narration can trigger a warning. He also describes code review by Codex, tests for identified issues and a human approval step.
Are the reported results independently verified?
No independent verification is provided in the account. The measurements and descriptions are attributed to Meyer and concern his own project.
What remains unknown about the system?
The account does not specify the gates’ thresholds, error rates or performance across a larger set of videos. It also does not establish whether the checks can reliably judge visual accuracy or overall audience quality.
Source: ThorstenMeyerAI.com
