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

# Conversation log: find and review every call

> Find any call your agents took: quick views, filters by status, channel, analysis and length, search, columns, one agent's calls, and exporting the log.

The Conversation log lists every call and chat your agents have handled in the project, newest first. Use it to find a call, narrow the list to the calls worth a look (not resolved, handed off, with errors), and open a call to read its transcript and analysis.

![The Conversation log on All calls, with the summary line, the quick views and the calls of today grouped by day](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/monitor/conversation-log-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=158606c8808bcf553ca228280c918180)

## Before you start

You need the **View** permission on the **Conversation log** ("See calls, transcripts and recordings."). Owners, Admins and Users have it by default. Exporting needs **Export**, and ending pending calls needs **Monitor (live)**, which Owners and Admins have by default. See [Roles and permissions](https://docs.tryhamsa.com/overview/projects/roles-and-permissions.md).

## Find the calls you want

**1. Open the log**

Click **Conversation log** in the sidebar.

**2. Pick a quick view**

The tabs above the list are quick views, each with its count. **All calls** shows every status. Use the others to jump to calls worth a look.

**3. Add filters**

Click **Filter** and pick what to narrow by. Each filter you add shows as a chip, for example "Date · is within · Last hour". Click a chip to change it, or **Clear** to remove every filter and the search.

![The Filter menu: Agent, Channel, Status, Date, Resolved, Handoff, Errors, Knowledge gap, Sentiment and Call length](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/monitor/conversation-log-filter-menu-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=6ff27b7ea9d3df854635ce736b7a8e96)

**4. Open a call**

Click a row to open its [call details](https://docs.tryhamsa.com/agents/monitor/call-details.md) beside the list.

### Quick views

| View | Shows |
| - | - |
| **All calls** | Every call, in every status. |
| **Live now** | Calls in progress, as a board of cards. See [Live calls](https://docs.tryhamsa.com/agents/monitor/live-calls.md). |
| **Pending** | Calls waiting to start, for example from a [batch call](https://docs.tryhamsa.com/agents/connect/batch-calls.md). |
| **Not resolved** | Calls the analysis judged not resolved. |
| **Handed off** | Calls handed to a person. |
| **Errors** | Calls where speech to text, the model or speech reported an error. |
| **Knowledge gaps** | Calls where the agent didn't have the answer. |
| **Negative** | Calls the analysis judged negative. |

The last five views use the AI analysis that runs after each call. [Analytics](https://docs.tryhamsa.com/agents/monitor/analytics.md#how-calls-are-scored) explains how it works. Use the arrow keys, Home and End to move between the tabs.

### Filters

| Filter | Options |
| - | - |
| **Status** | **Completed**, **Failed**, **Forwarded**, **Terminated**, **In progress**, **No answer**, **Pending**. Pick several; each shows its count. |
| **Channel** | **Telephone**, **Web**, **Chat agent**. Pick several. |
| **Agent** | One agent. |
| **Date** | **Last hour**, **Today**, **Yesterday**, **This week (from Sunday)**, **This month**, or a custom range on the calendar. The range applies once you pick both ends. |
| **Resolved** | **Resolved** or **Not resolved**. |
| **Handoff** | **Handed to a human** or **Not handed off**. |
| **Errors** | **Had an error** or **No errors**. |
| **Knowledge gap** | **Had a knowledge gap** or **No knowledge gap**. |
| **Sentiment** | **Positive**, **Neutral**, **Negative**. Pick several. |
| **Call length** | **Less than 30s**, **30-120s**, **More than 120s**. |

There is no "All" date option: to see every date, remove the date chip. Dates follow your browser's time zone.

### Search

Type in **Search by call ID, agent ID or phone number**. Search matches:

* a full call ID, exactly,
* a full agent ID, exactly,
* the caller's phone number, or any part of it.

It doesn't match the agent's own number, names or transcript text. Your other filters still apply to the results.

## Read the list

### The summary line

Above the list, one line sums up the calls on screen, for example "185 calls · 75% completed · 3 failed or unanswered · 64% resolved · 12 handed off · 0:52 avg length". Click a figure to list only those calls.

* **resolved** is out of the calls the analysis judged, not all calls. Hover it to see how many that is.
* **avg length** follows only the agent and date filters.
* On an agent with an [A/B test](https://docs.tryhamsa.com/agents/release/ab-tests.md) on prod, it adds how many calls came from the test.

### Columns

| Column | What it shows |
| - | - |
| Time | When the call started, with an icon for its status. Hover the icon for the status name. Calls in progress show a **Live** pill. |
| Agent | The agent that took the call. |
| **Caller** | The caller's number with its country flag, "Web visitor" for a website call, or "Chat". |
| **Version** | The agent version and [environment](https://docs.tryhamsa.com/agents/release/environments.md) that answered, for example "v3 · prod", or **Draft** for a test of a draft. |
| **Latency** | The call's average reply time in seconds. Amber when it is over 1.5 seconds. |
| **Analysis** | What the analysis found. "Analysing" while it runs, "Not analysed" if it couldn't run. |
| **Length** | The call's length like a clock (0:04, 5:11, 1:02:09), and how many turns it had. |
| **Credits** | Credits the call used. |

Hidden by default: **Timestamp**, **Agent number** and **User number**. Turn them on under **Display** > **Columns**.

Some rows carry flags: **Human took over** (a member [took over the call](https://docs.tryhamsa.com/agents/monitor/live-calls.md#take-over-a-call)), **Tool failed**, and **Slow replies** (average reply over 1.5 seconds).

### Sort, group and arrange

Open **Display** to change how the list looks:

* **Grouping**: **Day** (the default, while sorted by time), **Agent**, **Status** or **None**.
* **Ordering**: sort by time, **Length** or **Credits**, ascending or descending. You can also sort from a column header's menu. The default is newest first.
* **Columns**: show or hide columns.
* **Reset view**: back to the defaults.

![The Display menu with Grouping set to Day, Ordering by Time, the column chips and Reset view](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/monitor/conversation-log-display-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=1acdebedef491b379c4fd8af2a92b979)

Your display choices, the **Channel** filter and the rows per page are remembered for this project in this browser. Other filters and the search are not; they live in the page address, so you can bookmark or share a filtered list.

The list shows 10, 20 or 30 calls per page, with previous and next buttons.

### Refreshing

The list refreshes every 20 seconds, and when you come back to the browser tab. It stops while the tab is hidden. The length of a call in progress counts up every second.

## See one agent's calls

To see only one agent's calls, use one of these:

* In the agent's editor, open the header menu and choose **Calls** ("All calls for this agent").
* In the agents list, click the agent's call count or **View calls**.
* In the agent's releases, click **View prod calls** (or staging, dev).

The log opens filtered to that agent, with the agent's name and **Open agent** at the top. For agents with versions you get three more filters:

| Filter | What it does |
| - | - |
| **Environment** | prod, staging or dev. |
| **Version** | One or more versions, or **Draft** for test calls on a draft. |
| **A/B test** | **Only A/B test calls**: the calls an A/B test sent to a version. |

## Export the log

Click **Export** and choose one of these:

| Option | What you get |
| - | - |
| **Export current page** | An Excel file (.xlsx) of the rows on screen, downloaded at once. |
| **Export with filters** | Opens **Export conversation log**. Pick an agent (required), a date range (default **Today**) and a status (**All** or one), then click **Export**. The file is prepared in the background. When it's ready, you get an email with a download link, and it shows in your notifications. |

![The Export conversation log dialog with Agent, Date range set to Today and Status set to All](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/monitor/conversation-log-export-dialog-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=0d4d85822331116e9fc4dc391a4eae89)

The background export includes each call's analysis: resolved, handed to a human, sentiment, errors by stage, knowledge gap, and what was asked outside the agent's scope. It holds up to 1,000 calls, and the download link works for 24 hours. If your range has more calls, the export is refused; narrow the dates and try again.

The background export uses its own agent, dates and status. It doesn't follow the filters on the list.

## End calls that haven't started

Calls in **Pending** are waiting to start. They count against your project's limit on calls at the same time, so you may want to clear calls you no longer need.

* **One call**: on a pending row, click the end button and confirm.
* **Every pending call**: open the **Pending** view and click **End all N pending**. Confirm with **End N calls**. This ends every pending call that matches your filters. Calls in progress are not touched.

Ended calls get the status **Terminated** and never start. Pending rows don't open call details.

## Good to know

* The sidebar shows **Conversation log** and **Live calls** to every member. Without the **View** permission, the page says "You don't have access to this feature."
* Status and channel are filters, not columns. The status shows as an icon next to the time.
* Analysis filters and views only find calls the analysis has finished. A call that just ended can take a moment to show up in **Not resolved** or **Negative**.
* On a narrow window, each call shows on one line, and call details open over the list instead of beside it.

## Related

> [card] [Call details](https://docs.tryhamsa.com/agents/monitor/call-details.md)
>
> Transcript, recording, analysis, details and logs for one call.

> [card] [Live calls](https://docs.tryhamsa.com/agents/monitor/live-calls.md)
>
> Listen in, instruct the agent, or take over a call.

> [card] [Analytics](https://docs.tryhamsa.com/agents/monitor/analytics.md)
>
> The same calls as figures and trends.

> [card] [Webhooks](https://docs.tryhamsa.com/agents/connect/webhooks.md)
>
> Get call results in your own systems.


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