🔍 Read the full analysis: I Took An AI Idea To Production In One Evening—Here’s The Process on ThorstenMeyerAI.com
Get hardware and tech essentials 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 he built a hand-drawn video engine in one evening and published a short test video that night. The first full video the next morning was slow and visually sparse; after measuring the problems, he added checks that stop some weak scenes from being rendered.
Thorsten Meyer says he built a hand-drawn video engine in one evening, published a short test that night, then revised the system after its first full video looked sparse and slow. In an account published November 6 on ThorstenMeyerAI.com, he describes using a second AI to review code, human approval before publication and measured checks to catch problems in later renders.
Meyer says the idea came from a hand-drawn explainer video whose visual style he liked. To avoid copying it, he says he wrote down a limited set of broad traits: hand-drawn lines, a continuous visual world, and large numbers used as colour accents. He says the engine was built from scratch, without assets, layouts or drawing code taken from the reference. Its palette came from his own design system, and the handwriting font was openly licensed.
According to Meyer, Claude built the engine and OpenAI’s Codex reviewed it without contributing to the build. He says Codex identified five issues in the first review round, which he fixed that evening. Each fix, he writes, was paired with a test designed to fail against the old code. Meyer then watched a roughly one-minute test video and approved it for publication; it appeared that night in an article about local votes against data centers.
The first full video, made the next morning, did not meet his expectations. Meyer describes excess white space, icons that did not support the narration and drawings that stopped moving when complete. He says the team measured frame coverage and still-image time, then changed the engine. For the first three scenes, he reports, filled-frame coverage rose from 15% to 66%, time spent completely still fell from 29 seconds to zero, and camera moves increased from three to 27. He says the resulting checks now flag scenes that are too empty or include an object unrelated to what is being said.
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
From Video Demo to Production Checks
The account’s main point is not simply that an AI-assisted video can be built quickly. It shows the gap between a working demo and a result judged ready for publication: Meyer approved a short test, but the longer video exposed issues that were not apparent from whether the engine ran.
His reported measurements also show how subjective feedback can be turned into repeatable checks. Frame coverage and stillness are specific properties a system can flag, while a human still has to decide whether the imagery supports the narration. The process described is one creator’s account, not independent evidence that the same method will work for other teams or video systems.
Products Worth Considering
Logitech MX Keypad Programmable Macro Pad for Developers, Graphite
Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite
Apple MacBook Pro MPHF3LL/A 14.2-Inch Laptop | M2 Pro 12-Core Chip 16GB Unified Memory 1TB SSD | 19-Core GPU Liquid Retina XDR Display Space
As an affiliate, we earn on qualifying purchases.
How the First Render Changed
Meyer presents the episode as a personal case study in taking AI ideas into production, including failures. He says the initial design began with a reference video, making the boundary between inspiration and copying an early concern. His stated approach was to record general visual traits while creating original assets and code.
The revised video, he writes, went live a few hours after his critique of the first full version. He identifies it as a video about the Nasdaq’s worst day in 19 months. The original, weaker video remains online unchanged as an example of what the system produced before the redesign. The article also points to a separate experiment in which a model completed more legal-benchmark answers after a weight edit, but its legal score remained zero; Meyer uses that result to argue that completion counts alone do not establish quality.
““A demo that runs is not a decision.””
— Thorsten Meyer
What the Published Metrics Show
The reported figures come from Meyer’s own build notes and experiments, according to his article. The comparison covers the first three scenes of the rebuilt video; the account does not provide enough detail to establish how representative those scenes are of other projects, or how the metrics were calculated. It also does not report an independent evaluation of the videos or the new checks.
Meyer says the original reference video was not copied, but the article’s description alone does not provide an external assessment of that claim. The account does not specify the engine’s full technical architecture, operating costs, production time beyond the stated evening and following morning, or how often its new gates produce false alarms. The legal-model result is a separate experiment and does not establish that the video process improves model performance.
The Gates Will Test Future Renders
Meyer says the measurements have become checks in the build process: scenes judged too empty can be stopped before rendering, and objects that do not match the narration can trigger a warning. Future results would show whether those gates prevent the same problems across more videos, rather than only improving the example described in the article.
The source does not announce a release schedule, a broader product launch or an independent audit. For now, the published account and video document Meyer’s process and reported results; whether the checks generalize remains to be seen.
Key Questions
What did Thorsten Meyer build?
He says he built an engine for a hand-drawn video style that uses a continuous visual world, animated drawings and camera movement to accompany narration.
How did he say the system was reviewed?
Meyer says Claude built the engine and OpenAI’s Codex reviewed the code. He reports that five findings were fixed, with each fix paired with a test that failed against the earlier code.
What changed after the first full video?
After describing the first full video as sparse and slow, Meyer says he measured frame coverage and stillness and adjusted the engine. In the first three scenes, he reports coverage rising from 15% to 66%, stillness falling from 29 seconds to zero, and camera moves increasing from three to 27.
Are the reported results independently verified?
The article presents them as results from Meyer’s own notes and experiments. It does not provide an independent evaluation or enough methodological detail to judge how broadly the figures apply.
Source: ThorstenMeyerAI.com
