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

# Replay From Transcript

> Replay real conversations against your agent.

Use this type to test your agent again on conversations that went wrong, or on calls that must keep working after a change.

## Set it up

**1. Select the card**

[Create a simulation](/test/simulations/create.md#create-a-simulation) and select the **Replay From Transcript** card.

![Replay From Transcript card in the Create Simulation type picker](https://mintcdn.com/bluejay/EeUk0hD_moG9tWmI/images/create-simulation-card-replay-from-transcript.png?fit=max\&auto=format\&n=EeUk0hD_moG9tWmI\&q=85\&s=18853c2523c65f16213dc47d5b152de5)

**2. Add the transcripts**

Paste transcripts, select your agent's production calls, or upload a CSV file with the [CSV columns](#csv-columns) below. Each transcript becomes one Digital Human.

* Bluejay uses only the tab that is open when you click **Save**.

![The Direct Input tab with a pasted transcript and success criteria, the Add New Transcript button, and the Smart Replay, Strict Replay, and Persona Generation replay modes](https://mintcdn.com/bluejay/eKcFBkjZX47pI8Ss/images/replay-from-transcript.png?fit=max\&auto=format\&n=eKcFBkjZX47pI8Ss\&q=85\&s=ee6c268a4a44a95e5a8a35e9b0d834e9)

## Replay modes

| Mode | What the Digital Human says |
| - | - |
| **Smart Replay** | The transcript lines when they fit the conversation, and new lines when they do not. |
| **Strict Replay** | The exact lines of the transcript. |
| **Persona Generation** | New lines, based on the caller and the context of the transcript. |

Use **Strict Replay** when your agent must give the same answers as before. If your agent changed what it says, the fixed lines may stop making sense. **Smart Replay** is usually the better choice then.

## CSV columns

The CSV file has a header row and one transcript in each row. The columns are:

* `Transcript`
* `Name`, optional
* `Success Criteria`, optional
* `TC_<tool>`, optional, an expected tool call
* `CT_<trait>`, optional, the value of a [trait](/test/simulations/digital-humans.md#traits)

## Related

* [Logs](/monitor/logs.md)
* [Digital Human Reference](/test/simulations/digital-humans.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.