The run that was never resumed: 175 videos listed, 39 transcripts saved, rate limited three times and abandoned mid-backoff at video 40

Transcript Archive

CLIENT

Transcript Archive

YEAR

2025

INDUSTRY

Content research

SERVICES

AI Systems

Key Highlights

Dozens of hours of video, and no way to search it

Researching how a genre of build-in-public content talks about making and monetising software means watching hours of video. Transcripts let you read and search that material instead.

It stopped, and the number says how far

Of 175 videos found, 39 transcripts were saved before rate limiting blocked the run three times and it stopped mid-backoff. It was never resumed, so most of the catalogue was never collected. That is the whole story of this one.

A channel's transcripts, pulled for offline research

A script that lists every video on a channel and downloads each caption transcript to a local file, built against a build-in-public software channel.

Designed for measurable product growth

Every design decision focused on improving usability, increasing engagement.

175

Videos Found

The full catalogue on the target channel.

39

Transcripts Saved

Before rate limiting stopped the run for good.

3

Blocks

Backed off three times, then abandoned mid-backoff.

Works

Explore More Works

The dashboard drawn as an office floor, each named agent shown resting or paused, above a Run Pipeline bar set to a trade and a city
The scored inbox, each captured post given a score, a verdict of act, read or noise, and a next step, under a banner warning that the scores are uncalibrated and should be read as a hypothesis