> ## Documentation Index
> Fetch the complete documentation index at: https://api.thesapientcompany.com/llms.txt
> Use this file to discover all available pages before exploring further.

# How Sapient works

> What a brain-response scan measures, and how to think about the output.

Sapient is trained on real human fMRI brain responses. When you submit a piece of content, the model predicts how a human brain would respond to it, second by second, and distills that into a single, comparable read.

## Brain is the ground truth, AI is the explanation

The core idea is simple. The **brain response is the measurement**: the model predicts the actual pattern of activity a human brain would produce while watching, hearing, or reading your content. Everything above that is interpretation built on top of the measurement.

* The **brain response** is the ground truth: a predicted activation across the whole cortex, every second.
* The **lenses, scores, and grades** are supplementary. They translate that response into language a marketer or creator can act on.

So when a scan says "attention drops at second 6," that isn't a guess from a language model reading your script. It's a reduction of a predicted brain signal into a plain word.

## What a scan measures

Every completed scan returns:

* A per-second **timeline**: each of the four **lenses** (attention, memory, buy / not buy, manipulation) scored 0–100 for every second of the content.
* Detected **moments**: the standout seconds where a lens peaks or drops, so you don't have to eyeball the whole curve.
* A plain-English **summary** of what happened and where, in a sentence or two.
* The **raw** brain output: per-network activations, second by second, for teams that want to go underneath the lenses.

## How to think about it

A Sapient score is a **directional prediction**, not a guarantee. Use it to compare versions of the same idea (A vs B), to find the moment a piece loses attention, or to sanity-check creative before you spend. It is most powerful when you scan often and compare.

<Columns cols={2}>
  <Card title="How the model works" href="/knowledge-base/how-the-model-works">
    The full pipeline, from content to lenses, step by step.
  </Card>

  <Card title="The 4 lenses" href="/knowledge-base/lenses">
    What each lens tells you and when to use it.
  </Card>
</Columns>