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

# Chat History

> Access and inspect the complete message history of any Conversation to audit interactions and improve Agents.

Timely.ai's Chat History records every interaction — contact messages, AI Agent responses, internal Notes, and human support actions — in chronological order, preserved indefinitely in the Workspace.

## Message inspection

Opening a Conversation loads the complete history in the central column of the Inbox. Each entry displays the following parameters:

| Parameter           | Description                                                              |
| ------------------- | ------------------------------------------------------------------------ |
| **Sender**          | Contact (customer), active AI Agent, or responsible human attendant      |
| **Timestamp**       | Exact date and time of sending, in the Workspace timezone                |
| **Channel**         | Source channel icon (WhatsApp, Instagram, Widget, Telegram, Slack)       |
| **Content type**    | Text, image, audio, video, document, or internal Note                    |
| **Delivery status** | Sent, delivered, and read — available for channels that support receipts |
| **Active Agent**    | Badge indicating which AI Agent processed the message                    |

<Note>
  Internal Notes added by attendants appear in the history with a different visual highlight from messages sent to the contact. They are not visible to the customer.
</Note>

## Agent optimization

The Chat History is the primary source of information for improving Agent behavior. Two direct uses:

* **Identify Knowledge Base gaps** — messages where the Agent responded inaccurately or incompletely indicate topics that should be added to the Knowledge Base.
* **Review tone and language** — Conversations completed with a negative contact rating reveal response patterns that need adjustment in the system prompt.

To apply improvements:

* Copy problematic conversation excerpts and use them as negative examples in the Agent's system prompt.
* Create Q\&A entries in the Knowledge Base based on frequently asked questions identified in the history.
* Use Tag filters to segment Conversations by topic and review thematic batches.
* Export histories via the Conversations API for analysis in external BI tools.