
Why an 11-Second Loop Beats My Hero Video
I wanted a 35-second hero video. I built an 11-second loop instead: it loads faster, never shifts the layout, and lands the whole message in eleven seconds.
8 posts on this topic.

I wanted a 35-second hero video. I built an 11-second loop instead: it loads faster, never shifts the layout, and lands the whole message in eleven seconds.

FeedFoundry renders my Instagram text deterministically from templates, BuilderBob posts on its own and learns from real engagement back into planning.

Claude cuts my YouTube talks in the terminal: the agent proposes every cut, I approve, then it renders and uploads. One hundred percent local, no cloud.

waff Future Fit Festival, web3 hub vienna and Vienna Software Architecture Meetup: three appearances back to back, because I build, not the other way around.

A Whisper default replaced 13 minutes of my first real end-to-end transcript with a hallucination loop. The diagnosis, the fix and four transferable lessons.

Qwen3.6-35B technically loaded on an M4 Pro 24 GB at 100% GPU, yet ran at 0.1 tokens per second. A quiet failure, five lessons and hardware recommendations.

After two months and 1,052 commits, my autonomous infrastructure's AI moved from OpenClaw to Hermes Agent by Nous Research. The learning loop decided it.

At OpenClaw Hackathon Vienna, 150+ met at House of Innovation. Four people built an AI-powered B2B sales agent in one night, then made a video without sound.