Engine in testing · Claim intelligence

Epistell: an honest record of what people say in public.

Epistell listens to long form podcasts and video and writes down every claim: who made it, how firmly, and the exact second it was said, with a receipt that links back to the source. A clerk, not an oracle.

The problem

Analysts, journalists and investors spend hours listening to podcasts and interviews to find the one sentence that matters, then more hours proving it was said. Summaries lose the exact words. Transcripts bury them.

What the lab has delivered

  • Brand book and a locked logo, the E that writes itself
  • epistell.com, live with a waitlist for one hundred founding seats
  • Epistell Desk, a Mac app that runs a YouTube video or a whole playlist from start to finish
  • The engine: the creator's own captions or private transcription on the Mac, then every claim listed with its speaker and how firmly it was said
  • A check on every quote, word for word against the transcript, with times taken from the measured words and never from the model
  • The app shell: home, library, calls, people and the claim sheet with its receipt
  • A report library: ten formats in one house style, from a daily brief to a committee memo

What comes next

In order: capture from the phone, a test with listeners outside the lab, deeper analysis across sources, then the public record.

App screens use made up shows and people.

Related

Tell us where your week goes.

Tell us what eats your week, in plain words. We reply within two working days with a straight answer.

Request a build