About Verbatim
What Verbatim is, and how it works.
An AI's response is the sum of two things: information and the reasoning applied to that information. Both are fundamentally imperfect. An AI model (Claude Opus 4.7, GPT-5.5, etc.) compensates by writing with confidence, authority, coherence, and fluency — qualities that signal correctness without necessarily producing it.
The Verbatim Index is an attempt to correct for this by systematically separating the information layer from the reasoning layer, testing both under cross-examination. We run the same question through 9 frontier models, then have each one critique the others across four structured turns. What survives is verified. What gets challenged is disputed. What no model surfaces is a gap. The confidence, authority, coherence, fluency — stripped away.
A single-model output is unreliable in ways that we cannot detect from inside that output. The Index proves that with receipts.
The outcome
You make decisions with greater confidence knowing which parts of an AI's answer have held up under scrutiny and which haven't.
How it works
Verbatim applies cross-examination at three different weights, depending on what the situation requires.
The Debate.
A single competing model cross-examines an AI response in place. You're reading something in ChatGPT, Claude, Gemini, Grok, or Perplexity and want a second opinion. Verbatim runs one adversarial pass from a different model — critique, alternative framing, or distillation — without leaving the page. The everyday tool.
The Council.
Multiple competing models cross-examine the same response in parallel, then a synthesizer reconciles their critiques into a single verdict. You get the equivalent of a panel review in your browser. For when one second opinion isn't enough.
The Index.
The full methodology run on a fixed question with 9 source models, 8 critics per source, across four structured turns. Published as evidence. The Index proves at scale what the extension does at the point of consumption. The receipts.