cast_otter

Seoul · since 2010Numbers first

@cast_otter

← All works Works · 05

Channel & Blog Benchmark

Count 200 videos cheaply, then take apart only the 3 that matter.

Partly running5 key · Channel & Blog Benchmark

01The problem

Copying a channel that works “by feel” leaves no record of what was copied. Analysing everything in depth costs too much time and money. Count cheaply first, then spend the effort on a few.

02How it works · 3 steps

  1. Count

    Paste a channel and it pulls metrics for about 200 recent videos. No AI model is used at this step.

  2. Pick

    Two that did well and one that did average, so the difference shows.

  3. Take apart

    Only the 3 picks get the deep look, and what they teach goes into a playbook as patterns. The original content is never stored.

03Screens

The benchmarking work screen: step-by-step inputs and status cards, with the text blurred.
Real screen · data blurred
A channel analysis result: summary cards and a ranking table, with the content blurred.
Real screen · data blurred

04Stack

  • A video platform’s public API
  • A video-capable AI model, with a usage meter
  • Local Python and a web UI on a Mac
  • Built with an AI coding tool

05Status

Partly running

Channel analysis is running. Blog analysis is still at the design stage, so everything on the screen is the channel side.