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

# Using connectors in chat

> How the agent puts your connected tools to work mid-conversation.

Once a tool is connected and allowed, you don't do anything special to use it — you just ask. The agent reaches into the connected tool as part of answering you.

## Asking a question

Mention the tool, or just ask naturally — the agent knows what you've connected:

<CodeGroup>
  ```text HubSpot theme={null}
  How many deals are in our pipeline right now?
  What's the recent activity on Acme Health?
  ```

  ```text Power BI theme={null}
  Pull the patient-volume figure from our operations dashboard.
  ```

  ```text Meta Ads theme={null}
  Which ad campaigns produced leads last month, and at what cost?
  ```
</CodeGroup>

The agent reads the live data, summarizes it, and folds the answer into the conversation — often into a [report](/workspace/reports-and-artifacts) alongside your other numbers.

## Building a list to reach

You can start outreach directly from a connected tool — no export, no re-upload:

```text theme={null}
Call our HubSpot contacts in the "needs follow-up" stage.
```

The agent pulls the list from the connector and takes you into the normal [launch flow](/workspace/launching-outreach) — with the verified count and the approval card. As always, nothing dials until you approve.

## Writing back

When you ask the agent to change something in a connected tool — log a call, add a note, create a task — that's a **write**. Writes are set to ask for approval by default, so the agent shows you an [approval card](/connectors/permissions#approving-a-write-in-chat) previewing the exact change first.

```text theme={null}
Log this call outcome on the contact in HubSpot.
Create a follow-up task for next Tuesday.
```

## When something isn't available

* **Not connected yet** — the agent tells you the tool isn't connected and points you to **Settings → Connectors**.
* **Blocked or needs approval** — the agent respects your [permission settings](/connectors/permissions). A blocked capability is never used; a write pauses for your approval.

## Next

<CardGroup cols={2}>
  <Card title="Permissions" icon="sliders" href="/connectors/permissions">
    Control what the agent can do with each tool.
  </Card>

  <Card title="Launching outreach" icon="phone-arrow-up-right" href="/workspace/launching-outreach">
    Turn a connector list into a run.
  </Card>
</CardGroup>