> ## Documentation Index
> Fetch the complete documentation index at: https://docs.wittify.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Agent Overview
> The default home page for any single agent. KPIs, charts, intents, sentiments, ratings, escalations, recent conversations, and an activity log. Filter the date range and the entire page recomputes.
Open this page by clicking any agent in the **All Agents Overview** panel on the dashboard, the **My Agents** sidebar list, or the in-sidebar **agent switcher**. The route is `/agents/{agentId}`. This is the landing tab for Manage Agent and the default destination of every agent link in the app.
## How this page works
Every chart, KPI, and table on the page is **filter-reactive**. Pick a date range from the hero (Last 7 days / Last 30 days / Last 90 days / Custom) and every section recomputes against that window. There is no Apply button, the page updates as you change the range. The selected range is shared across the whole page, you don't pick it twice.
## Manage Agent shell
Before the content, the surrounding chrome:
### Sidebar nav
When you're on a Manage Agent route, the dashboard sidebar swaps in a per-agent navigation. From top to bottom:
* **Back to Home** , returns you to the main dashboard active view.
* **Agent switcher** dropdown , lists every agent you can switch to, plus a **Create an Agent** entry at the bottom that opens a 3-card type picker (**Text Agent**, **Voice Agent**, **Hybrid Agent**) and routes straight into the Create Agent wizard for the variant you pick.
* **Main items**:
* **Agent Overview** (the page you're reading, highlighted)
* **Chat Conversations** , hidden for voice-only agents
* **Voice Conversations** , hidden for text-only agents
* **Leads / Tickets**
* **WhatsApp Campaigns** (text + hybrid only), **Outbound Voice Campaigns** (voice + hybrid only). Both groups have an **Archived** + **Deleted** chevron toggle.
* **Agent Settings** collapsible , deep-links into each wizard section as an inline editor: **Text Settings** / **Voice Settings** / **Appearance** / **Knowledge Base** / **Tickets & Leads** / **Integrations** / **Deployment**. Auto-expands when you're on a settings sub-route.
* **Assigned Users** collapsible , placeholder rows for upcoming team features (every item shows a **Coming Soon** chip).
### Topbar
The standard dashboard topbar with two additions on this surface:
* **Test Your Agent** button on the end edge, present on every Manage Agent page. Opens the test sheet documented at [Test Your Agent](/en/Systems/AI-Agents/Manage-Agent/Test-Your-Agent).
* A breadcrumb that reads **Dashboard / Agents / `{agent name}`**.
## Hero with date range
The first section of the page. Shows:
* **Agent name** in a large heading
* **Type chip** , one of **Text** (blue), **Voice** (indigo), **Hybrid** (purple)
* **Status chip** , **Active** (green) or **Draft** (muted)
* **Date range selector** on the end edge, four buttons:
| Range | What it covers |
| ---------------- | ----------------------------------------------- |
| **Last 7 days** | Most recent week |
| **Last 30 days** | Default. Most recent 4 weeks. |
| **Last 90 days** | Most recent quarter |
| **Custom** | Opens a calendar popover for any From / To pair |
The Custom popover has two month-views side by side, click a start date then an end date to set the range. Click **Apply** to commit. Custom dates display in your locale (Arabic Saudi Arabia in Arabic mode, otherwise English).
## KPI strip (5 cards)
A 5-column grid (collapses to 2 + 3 on smaller screens). Each card shows a title, a number, a delta vs. the previous window of the same length (green up-arrow / red down-arrow), and a small sparkline at the bottom.
| Card | What it counts |
| --------------------- | ------------------------------------------------------ |
| **Conversations** | Total chat + call sessions in the range |
| **Avg Response Time** | Mean time from a customer message to the agent's reply |
| **Tickets Created** | New support tickets the agent filed |
| **Leads Captured** | New leads the agent collected via the lead form |
| **Avg Rating** | Average post-conversation rating, on a 5-star scale |
When a delta is calculated against zero (the agent had no prior data), the card shows a **New** chip instead of a percentage.
## Channel chart
A donut chart titled **Channels** showing where conversations came from in the date range:
* **Web** (indigo)
* **WhatsApp** (green)
* **Messenger** (blue)
* **Instagram** (gradient pink)
* **X** (black / white)
* **Telegram** (sky blue)
* **Phone** (slate, voice only)
Hover any slice for a tooltip with channel name and percentage. The legend on the side lists every channel with its count.
## Day chart
Bar chart titled **Conversations by Day**. X axis is days of the week (Sunday-first per product direction in Arabic, Monday-first in English). Y axis is conversation count.
## Hour chart
Bar chart titled **Conversations by Hour**. X axis is 0 to 23. Y axis is conversation count. Useful for spotting peak hours and staffing your phone line accordingly.
## Widget Funnel
5-stage funnel showing how many visitors made it through each step of the chat:
1. **Opened** , widget appeared on the page
2. **Started** , visitor clicked into the chat
3. **First message** , visitor sent at least one message
4. **Resolved** , conversation reached a resolved state (positive sentiment + no handoff)
5. **Rated** , visitor rated the conversation
Each step's bar shows count and conversion percentage from the previous step.
## Voice metrics card (voice + hybrid only)
A grid of voice-specific KPIs:
* **Avg Call Duration** , mean call length
* **Peak Hour** , the busiest single hour in the range
* **AHT** (average handle time) , total call time including hold and post-call work
* **Call Drops** , calls that ended abnormally (caller hung up before agent responded)
Hidden for text-only agents.
## Text metrics card (text + hybrid only)
A grid of text-specific KPIs:
* **Avg Response Latency** , how long from customer message to agent reply
* **Messages per Conversation** , mean message count
* **Character Counts** , total inbound + outbound character count
Hidden for voice-only agents.
## Resolved card
A donut showing the split between **Resolved** and **Unresolved** conversations. The center number is the resolved percentage.
## Sentiment card
A stacked horizontal bar with three segments: **Positive** (green), **Neutral** (slate), **Negative** (red). Hovering shows the count for each segment.
## Ratings card
A 5-row chart showing rating distribution. Each row is one star count (1 through 5) with a horizontal bar showing how many conversations received that rating. The summary at the top shows the overall mean (e.g. **4.3 of 5** with **152 ratings**).
## Intents card
The most common reasons customers contacted the agent. Each intent row shows:
* The intent name
* A horizontal bar with the count
* A delta vs. the previous window
A **Customize Intents** button on the corner opens a modal:
### Customize Intents modal
Title: **Customize Intents**. Description: *Toggle the built-in intents your agent tracks, and add custom ones for your business.*
The modal has a draft pattern (your changes don't persist until you click **Save**):
* **Built-in intents** (toggle each on or off via switch). The active count is shown in the title (e.g. *14 of 18 intents enabled*).
* **Custom intents** (rows you've added). Each has a delete trash icon. Removing a custom intent opens a confirmation: *Remove "`{intent name}`"? This cannot be undone.* Built-in intents cannot be deleted, only switched off.
* **Add custom intent** input at the bottom. Up to **60 characters**. Case-insensitive deduplication, you can't add **Refund** if **refund** already exists.
Two buttons in the footer: **Cancel** (discards) and **Save Changes** (persists).
## Language card
Distribution of conversation languages (English, Arabic, plus any others detected). Each row shows language name, count, and a percentage bar.
## Escalations card
How many conversations escalated to a human. Two numbers: total escalations and percentage of conversations escalated. A small text under: *Escalation = a conversation where the Human Handoff tool fired.*
## Recent Conversations
A scrollable list of the most recent conversations in the range.
### Desktop layout (7-column grid)
| Column | Content |
| --------- | --------------------------------------- |
| `#number` | Conversation ID |
| Time | Relative timestamp (e.g. *2 hours ago*) |
| Channel | Brand icon dot |
| Sentiment | Pill (positive / neutral / negative) |
| Status | Pill (open / resolved / handoff) |
| Duration | e.g. *3:42* |
| Summary | First line of the conversation |
### Mobile layout (stacked card)
A two-row card per conversation: ID + time on the top row, channel + sentiment + status pills on the next, the summary as a line below. Tapping the card opens the same destination as desktop.
Click any row to deep-link into the [Chat Conversations](/en/Systems/AI-Agents/Manage-Agent/Chat-Conversations) inbox with `?selected=` so the conversation is pre-selected when the inbox opens.
## Activity Log
A paginated event feed of agent-level events: **Agent activated**, **Settings saved**, **Integration connected**, **Agent deactivated**, **Knowledge file uploaded**, etc.
Each row has an icon, a one-line label, and a timestamp. Pagination controls at the bottom: **Prev** / **Next** with **Page X of Y**.
## Variant differences (recap)
* **Text-only agents** hide: Voice Conversations sidebar item, voice settings deep-link, voice KPIs in the strip, the Voice metrics card.
* **Voice-only agents** hide: Chat Conversations sidebar item, text settings deep-link, text KPIs in the strip, the Text metrics card, social channel slices in the Channel chart.
* **Hybrid agents** see everything.
## Common questions
That's expected. Until the agent receives traffic on at least one connected channel, every chart is empty. Activate the agent in [Deployment](/en/Systems/AI-Agents/Wizard/Deployment), connect a channel, and counts populate as conversations come in.
Recent Conversations always shows the latest conversations in the date range, but it caps at the most recent 50. If your agent had less than 50 in the wider range, the list won't grow when you stretch it.
Sentiment is computed from message tone (the model's read of how the customer felt). Ratings are explicit stars the customer gave at the end. They can disagree, customers often rate above their actual sentiment out of politeness, or below it after a bad final touchpoint. Use both as signals.
Not from this page directly. The Tickets & Leads page has CSV export, the Conversation pages can export individual transcripts, and a full data export is on the roadmap. For now, screenshot the dashboard for sharing.
Numbers are computed on every page load. There's no live polling, refresh the page to see the latest.
## Next
The full chat inbox for this agent.
Recordings, transcripts, and AI summaries.
Captured records grid with filters and CSV export.
Edit any wizard section without leaving Manage Agent.