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

# What Are Rehearsals Replicas?

> AI replicas of real customers, how they differ from synthetic personas and a general AI, and why that difference matters

## What Rehearsals Does

Rehearsals builds AI replicas of real people, modeled from structured behavioral interviews,
social media, and other 3rd party data. You put a scenario in front of hundreds of your actual
customers and see what they would do, in minutes instead of weeks.

<Note>
  A replica is an AI representation of a specific, real individual. It encodes that person's
  decision-making patterns, constraints, values, and reasoning style, rather than a demographic
  average. Replicas are powered by the Engram Model, our proprietary model for replicating
  individual human behavior.
</Note>

## How Rehearsals Is Different

Other tools predict customer behavior by reasoning about an "average" consumer, a general profile
built from aggregate data. But real people do not act like an average. Their choices depend on
their own circumstances, habits, and priorities, which is exactly what an average washes out.

Rehearsals works the other way around. Instead of one generic consumer, it asks hundreds of
replicas of your real customers what each of them would do, then adds up the results. You see the
real range of responses, not a single averaged guess.

| Approach                                                    | How it works                                     | Where it falls short                                                             |
| :---------------------------------------------------------- | :----------------------------------------------- | :------------------------------------------------------------------------------- |
| Asking a general AI (ChatGPT/Gemini)                        | Guesses what a typical consumer would do         | Falls back on the average and misses the real range                              |
| Synthetic personas ("You are a 34-year-old man from Utah…") | Role-plays a demographic stereotype              | Speaks for a whole group, cannot say why, and is easily talked out of its answer |
| Rehearsals replicas                                         | Asks replicas modeled on real, individual people | Keeps the real spread of opinion, including the surprising responses             |

<Info>
  What sets a replica apart is that it can tell you why. Say a streaming service raises its
  subscription price by a few dollars a month. A synthetic persona might report that some
  percentage of a demographic would cancel, but not the reason. A replica reasons from a real
  person's life: "My kids watch this every day, and a few dollars a month is not worth the hassle
  of switching." Add up hundreds of those individual reasons and you get a realistic picture of who
  stays and who leaves.
</Info>

## Where to Go Next

<CardGroup cols={2}>
  <Card title="How Replicas Are Built" icon="dna" href="/docs/methodology/how-replicas-are-built">
    The Engram Model, the data behind each replica, and the four-stage construction pipeline.
  </Card>

  <Card title="What Is a Simulation?" icon="flask" href="/docs/methodology/what-is-a-simulation">
    The scenario, the environment, and how a replica is interviewed.
  </Card>

  <Card title="Simulation Types & Templates" icon="layer-group" href="/docs/methodology/simulation-types">
    Pricing, product preference, advertising, and the templates you can start from.
  </Card>

  <Card title="Asking Good Questions" icon="circle-question" href="/docs/methodology/asking-good-questions">
    How to write neutral questions, and why framing matters even more for AI.
  </Card>
</CardGroup>