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

# 02. The cabinet

> Where things are: what lives at the agent level and what lives at the organization level.

# The cabinet: where things are

> [!NOTE]
>
> **Module goal:** stop hunting for sections and understand what lives at the agent level and what lives at the organization level.

![The Overview dashboard: cards for “now”, volume, outcomes and agents](https://mintcdn.com/hubtalk/8xTC4yxNv5HnlhMA/images/v4_dashboard_overview_en.png?fit=max\&auto=format\&n=8xTC4yxNv5HnlhMA\&q=85\&s=a922ce6056717831e4d5abbef8e2d2a4)

/dashboard — the Overview dashboard: a header with the period and an agent filter, cards for “now”, volume, outcomes and agents.

The **Dashboard** section is built from cards: how many calls there were over the period, how many reached a conversation, how they ended, what is happening right now. The organization gets a ready-made **Overview** dashboard; you can edit it and create your own next to it. **Recent agents** and **Recent sessions** are two cards from the catalog. In detail — in [Dashboard](https://docs.hubtalk.ai/v4/platform/dashboard.md).

> [!NOTE]
>
> **Dashboard figures count live telephony only.** Chats, website calls and your test runs are not included — so **Calls** on the dashboard and the number of rows in History do not have to match.

The sections on the left are arranged in six groups, and this splits the areas of responsibility. This is how the menu looks to an organization administrator:

| Group | Sections | What matters |
| - | - | - |
| **Workspace** | Dashboard, Agents, Telephony, Channels, Campaigns, Calling rules (only in a `us_tcpa` organization), Do not call, History | Daily work: agents, who calls whom and what came of it. |
| **Platform** | Incidents, API keys | Observer signals and access for integrations. |
| **Catalog** | MCP servers | External tools that agents refer to. |
| **Library** | Components, Knowledge Bases, Phrase fixups, Voicemail phrases, Call screener phrases | Reusable parts of a conversation. |
| **Help** | Documentation | This guide. |
| **Admin** | Organization, Settings | The organization's users and the cabinet settings. |

Other roles see part of this menu: sections the role has no rights to do not appear in it. Which sections each role can open — in [Roles and access](https://docs.hubtalk.ai/v4/platform/roles.md).

MCP servers in the **Catalog** group are shared across the whole organization: an agent only refers to the record. So a server is edited in one place, not in every agent.

> [!NOTE]
>
> **The organization's market: `kz` and `us_tcpa`**
>
> The set of sections and the rules a campaign uses to choose the time of a call depend on the organization's market profile.
>
> * **`us_tcpa` (USA).** The **Workspace** group has a **Calling rules** section. It opens the **Calling rules (US)** screen — every role in the organization sees it, and it is read-only: the federal rules, the legal calling windows for 56 jurisdictions (the states, the District of Columbia and the territories), restricted days and holidays, legal frequency caps and the **Forbidden-windows grid**. A contact's location is determined by ZIP: for a contact with a ZIP the window is counted in a single time zone, and without a ZIP dialing is allowed only when the time in every candidate zone of the state is inside the window. A campaign's **Working hours** apply inside the legal window. In detail — in [US campaigns](https://docs.hubtalk.ai/v4/platform/campaigns/us-market.md).
> * **`kz` (Kazakhstan).** There is no **Calling rules** section in the menu. When to call is set by the campaign itself: **Working hours** on the **Schedule** step of the campaign wizard. If working hours are not set, dialing is limited to the organization's calling window — 08:00—21:00 by default.

![The agent list with filters and publishing status](https://mintcdn.com/hubtalk/UuNkjJ26eB-xpPwv/images/tutorial/tut-03-agents-list_en.webp?fit=max\&auto=format\&n=UuNkjJ26eB-xpPwv\&q=85\&s=20c1d8ccd11b1e341bf83f3d3fd4efba)

/agents — the agent list: folders on the left, a table with visibility, publishing status and modification date.

The **Agents** section is a table (**Name · Visible to members · Status · Last modified**), and on the left is a collapsible **Folders** panel with an **All agents** node. A folder is created with the **New folder** button; an agent is moved with the **Move to** item in the **⋯** menu of its row.

> [!NOTE]
>
> **A folder is a way of arranging the workspace, not a property of the agent.** Moving does not create a flow version and does not change edit attribution. Deleting a folder moves its agents to **No folder** and **does not delete any of them** — the confirmation dialog says so explicitly. An agent created while a folder is selected (and a duplicate of an agent from a folder) ends up in that folder.

An empty list clearly separates two cases: “the organization has no agents yet” and “nothing matched your search or filter”.

The **Visible to members** column shows whether members see the agent. It is changed with the **Visibility** item in the row's **⋯** menu, and the **Filters** menu has a visibility filter. A hidden agent keeps calling, but members are not offered it for new campaigns. The same switch is in the builder: **Settings → Additional → Access**, **Hide from members**. Why this is needed and what exactly is hidden — in [Agents](https://docs.hubtalk.ai/v4/platform/agents/list.md).

![The agent creation dialog](https://mintcdn.com/hubtalk/UuNkjJ26eB-xpPwv/images/tutorial/tut-04-agent-create_en.webp?fit=max\&auto=format\&n=UuNkjJ26eB-xpPwv\&q=85\&s=016b565fce4ba13c8610ae16dbeb00ff)

Creating: the New agent button. A new agent is created with the minimal Begin → Conversation → End template. Use Latin characters for the name: it goes into the URL and into the call history.

**Duplicate** also asks for a name: the dialog opens with the first free variant. If you enter a name that is taken, the system refuses: the field is highlighted, and below it appears “An agent named «X» already exists”. The name of an **archived** agent also counts as taken — only deleting the agent frees it. After copying, the notification “Agent “X” copied” appears.

> [!TIP]
>
> **Practice**
>
> Create an agent `firstname_lastname_ucheba` and open it. All the following modules are done in it — repeating what the screenshots of the training agent `amanat_zapis` show.


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