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

# Chat History

> Every written conversation your receptionist had, and what she said about her own performance.

**Chat History** is where you read what actually happened. Every written conversation is here, with your receptionist's own assessment of how it went.

This is the screen that closes the loop: she reports what she could not answer, you add it to [Training](/ai-manager/training/overview), and the gap stops happening.

***

## What is on the screen

A list of conversations, newest first. Open one and you see the full exchange, plus:

* **How she graded herself** against the [quality criteria](/ai-manager/receptionist/chat) you set
* **What was missing** when she fell short — the topic, not just a flag
* **What the guest thought**, where they told her

Conversations she handled cleanly are marked as such. The ones worth your time are the ones that were not.

***

## What you do here

<Steps>
  <Step title="Filter to what needs attention">
    You are not here to read every conversation. You are here for the ones she flagged.
  </Step>

  <Step title="Read the exchange, not just the verdict">
    The transcript usually shows you *how* the guest asked, which is often why she missed it — guests rarely use your internal words for things.
  </Step>

  <Step title="Decide what it means">
    Most flags mean a document is missing. Some mean a document exists but does not use the words a guest would. A few mean the guest asked something no hotel could answer.
  </Step>

  <Step title="Fix the document, not the conversation">
    The conversation is over. The document is what stops the next one going the same way.
  </Step>
</Steps>

<Tip>
  **Read this weekly, not daily.** A week gives you enough conversations to see a pattern. The same missing topic three times is a much clearer instruction than one flag on a Tuesday.
</Tip>

***

## What happens next

Nothing changes from this screen. It is where you decide what to change somewhere else — almost always a document in [Training](/ai-manager/training/overview). Once that document is in, she can answer the question the next time it comes up.