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

# Conversation Intelligence: the issues customers raise with your AI agent

> What Analytics → Conversation Intelligence shows: AI replies, conversations and the issues customers raise with your WhatsApp and Onsite AI agent, with filters by date and sentiment and the conversations behind each issue.

**Conversation Intelligence** shows the issues that customers raise in their chats with your AI agent. It groups these issues by type and shows how often each type comes up. Open it in **Analytics → Conversation Intelligence**, or go to [https://ai.textyess.com/conversation-intelligence](https://ai.textyess.com/conversation-intelligence).

It reads the chats that your AI agent answers on WhatsApp and with the Onsite Agent.

## When Conversation Intelligence is available

**Conversation Intelligence** is in the sidebar under **Analytics**.

The page works when your plan includes WhatsApp or the Onsite Agent. If your plan includes neither, the page shows **Unlock Conversation Intelligence** and a **View Plans** button. You find your plan on the **Plan** page: [https://ai.textyess.com/organization-settings/plan](https://ai.textyess.com/organization-settings/plan).

If the page shows "No intelligence data yet", TextYess found no issues in the selected dates. Choose a wider date range. If your AI agent is off, turn it on. See [WhatsApp agent](/agents/whatsapp-agent) and [Onsite agent](/agents/onsite-agent).

## What Conversation Intelligence shows

At the top of **Analytics → Conversation Intelligence**, three cards cover the selected dates:

* **Total AI replies**: the messages that your AI agents sent on WhatsApp and onsite.
* **Total conversations**: your WhatsApp conversations. A conversation ends after 24 hours with no message.
* **Issues detected**: how many issue types TextYess found.

Under the cards, a table lists each issue type:

* **Issue**: the name of the issue type.
* **Users**: how many customers raised it.
* **Interactions**: how many customer messages, each with the AI reply, belong to it.

The issue with the most interactions comes first. Click an issue to read the chats behind it.

## Filter Conversation Intelligence by date and sentiment

At the top right of **Analytics → Conversation Intelligence**:

1. Choose the dates: **All**, **24h**, **7d**, **30d** or **Custom**. The page opens on **30d**.
2. Choose the sentiment: **All sentiments**, **Positive**, **Neutral**, **Negative** or **Very negative**.

**24h** shows yesterday. **7d** and **30d** end yesterday. **All** covers the last 2 years. For **Custom**, click the first day and then the last day in the calendar.

The sentiment filter keeps only the messages with that sentiment. Choose **Negative** or **Very negative** to see the issues that upset customers most. When you open an issue, the issue page keeps your dates and your sentiment.

## Read the conversations behind an issue

1. Open **Analytics → Conversation Intelligence**.
2. Click an issue in the table. A page opens with the title **Issue:** and the issue name.
3. Click a row to open **Interaction details**.

The issue page lists each interaction, newest first:

* **User query**: what the customer asked.
* **Date**: when it happened.
* **From**: the customer's phone number, or **Onsite Chat** for the Onsite Agent. Click the phone number to find the chat in the Inbox. Click **Onsite Chat** to open the onsite chat.

**Interaction details** shows the date, the channel, the phone number, the customer's message and the AI reply. For a WhatsApp chat, it also shows the customer's details. Click **Back** to return to the issue list.

## Export the conversations of an issue to CSV

1. Open **Analytics → Conversation Intelligence**.
2. Click an issue in the table.
3. Click **Export to CSV**.

The message "Export started" shows, and the download starts. The file has all the interactions of that issue for your dates and sentiment. The columns are the customer message, the AI reply, the channel, the phone number, the onsite chat ID, the time and the user query.

If the message "Export failed" shows, try again. The button shows only when the issue has interactions.

## Does Conversation Intelligence summarize the topics customers write about?

Only in part. **Analytics → Conversation Intelligence** lists the issues that customers raise with your AI agent, sorted by how often they come up. It shows problems, not every topic that customers write about. It does not write a text summary of your chats.

It reads only the chats that your AI agent answered. Chats that only your team answered are not in it. Test chats and the preview of your agent are not in it either.

To read the chats behind one issue, click the issue in the table. To see the most common problems in a period, choose the dates and read the top of the table.

## How fresh Conversation Intelligence data is

**Analytics → Conversation Intelligence** is not live:

* The issue list for a date range can be up to 24 hours old.
* **Total AI replies** for WhatsApp updates once a day.
* **24h**, **7d** and **30d** end yesterday, so today is never in them.

If a recent chat is missing, check the page again the next day.


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