Case Study· 2026· Internal

Transcript Desk, video research

Transcript Desk turns video into research material: a link becomes a timestamped transcript, a summary and a table of claims worth checking. It runs entirely on your own computer — nothing goes to the cloud, no subscription. The code is public on GitHub.

See the full portfolio
Client
Internal Project
Industry
Research & content
Platform
Local, on your own Mac
Year
2026
Status
Open source
The challenge

An hour of video for a minute of information

More and more primary research lives in video and podcasts: interviews, talks, hour-long discussions you need three claims from. You scrub, take notes by hand, lose the timestamp, and the transcript — if it exists — is locked inside the platform holding the video. The tools that do this in the cloud want a subscription and get to see all your research material.

The solution

Link in, working material out

We built a local desk: you give it a link, and the app pulls the audio, transcribes it with timestamps, and produces a summary plus a claims table — what was said, when, and what deserves checking. Everything runs on your own Mac, so the research material never leaves the machine. We built it for our own research, then published it on GitHub.

Visual identity

Color palette

A dark desk with warm off-white text and almost no color — coral shows up only for errors. The transcript is the star, not the interface.

Graphite
#16181D
Panel
#1B1E24
Bone
#E4E2DD
Stone
#97948D
Coral
#FF8078
Paper
#FCFBF8
Typography
Primary display
SF Mono

A monospace-led desk — timestamps, transcript and the claims table align to the character.

Secondary body

System sans for controls and labels, monospace for everything that is working material.

Monospace for material, sans for controls.

Process

From research to launch

01
We started from our own workflow: what we actually look for in an hour of video, and in what form we need it.
02
We built the local pipeline: pull the audio, transcribe with timestamps, no cloud services.
03
We added the research layer: a summary and a claims table, each pointing back to the exact moment.
04
We packaged it as a simple local app, started with one click.
05
We published the code on GitHub, so anyone can run it on their own machine.
Information architecture
  • Input
    • Video or podcast link
  • Transcript
    • Timestamps on every line
  • Research
    • Summary
    • Claims table
The app

From link to working material

Fact, opinion, speculation

Any note can get an on-demand research pack: the exact quotes from the transcript, split into fact, opinion, and speculation, plus verification queries ready to search. Every claim keeps the second it was said, so checking starts at the source.

The reading list

Books get their own section: add a title and the author, description, and tags fill themselves in. From a note you've already captured, whether a BookTok video, a tweet, or an article, the app pulls every book it mentions and keeps the source's own framing in the first tag. The list exports as Markdown.

Results

100%
processing on your own Mac
Public
open code on GitHub
Your project

The next case study could be yours.

Tell us what you want to build, and we'll turn your idea into the next project in our portfolio.

Start a project
More work

Other projects

See all case studiesSee all