> ## Documentation Index > Fetch the complete documentation index at: https://docs.moneypenny.ai/llms.txt > Use this file to discover all available pages before exploring further. # Calls & Messages > Review the conversations your agents handled in any period — who got in touch, how it went, and the full transcript, summary and details collected. **Calls & Messages** (under **Activity** in the sidebar) lists the conversations your agents have handled in a period, calls and chats alike, newest first. Click any one to read exactly what was said. For totals and trends across those conversations, use [Analytics](/monitoring/analytics) — the two pages share the same filters and count conversations the same way, so the conversations listed here are the ones behind your Analytics figures, and the numbers always agree. ## Choose what you're looking at The filter row at the top works exactly as it does on Analytics: * **Timeframe** — **Day**, **Week** or **Month**, with arrows to step back through earlier periods, or **Custom** for a range of your own (up to one month), with a from and to date and time. Until you've picked both dates, the page asks you to choose a date range. * **Agent** — one agent, or **All agents** for the whole workspace. * **Sentiment** — **Any sentiment**, or only the conversations graded a particular way, from **Happy** through **Neutral** to **Angry**. The colours match the **How callers felt** bar on Analytics, and clicking a segment of that bar brings you here with the sentiment already chosen. * **Timezone** — the clock the times are read against. If your account offers more than one zone you can switch it, and your choice is remembered. A line under the filters always states exactly what's on screen — the period, the agent, the timezone, and that you're looking at completed conversations, excluding test chats. When you're on the current period it also says **live**, because that period is still filling up. ## The conversation list Above the list a count tells you how many conversations match — "12 conversations · most recent first", or "3 angry conversations · most recent first" while a sentiment filter is on. Each row is one conversation: * **Time** — when it ended. * **Caller** — the number that got in touch, where there was one. * **Duration** — how long it lasted. * **Sentiment** — how the caller came across, or **Not analysed** if the conversation wasn't graded. * **Unanswered** — a **Question** chip when the caller asked something your agent couldn't answer. A dash means the conversation wasn't analysed. * **Channel** — **Phone**, **SMS** or **Chat**. * **Agent** — which of your agents handled it. The list is paginated. If nothing matches, the page says so and suggests stepping back with the arrows, widening the timeframe to a week or a month, or — under a sentiment filter — clearing it. ## Review a conversation Click a row and the conversation opens in a panel on the right, headed with the day and time it ended. Beneath that are the caller (or **No caller recorded**), the channel, the duration and the agent, then the sentiment and — if the conversation was flagged — a **Spam / wrong number** chip. Below, in order: * **Couldn't answer** — appears only when the caller asked something your agent couldn't answer, and quotes each question. When you see one, the fix is usually to add the answer to your [knowledge library](/knowledge/knowledge-library). * **Summary** — a short written summary of what the conversation was about and how it went. * **Transcript** — the whole conversation, turn by turn, labelled **Customer** and **Agent**. This is the definitive record of what was actually said. * **Details collected** — the information your agent captured during the conversation — a caller's name, their number, the reason they got in touch — as a simple field-and-value table. Close the panel with the **×** in its corner to get back to the list. A conversation is listed once it has ended, so a call that's happening right now appears here — and in your Analytics figures — when it finishes. Test conversations held on an agent's [test drawer](/building/testing) are never listed; you review those from the drawer itself, which opens the same panel.