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

# AI agents

> Connect StageWhisper to AI assistants that can act on what happens during your calls.

StageWhisper can connect to AI assistants that act on what comes up during your calls. Instead of manually creating follow-up tasks afterward, the AI picks up on actionable moments in the conversation and routes them to your assistant.

Supported AI assistants include [OpenClaw](/agents/openclaw) and Hermes (coming soon). StageWhisper speaks a [BYO protocol](/agents/byo-protocol) that any AI assistant can implement. If you'd rather not connect anything, you can [run a model on your own device](/agents/local-model) instead.

## What this looks like in practice

During a call, StageWhisper might detect that a prospect asked for a case study. If you have an AI assistant connected, it creates a task: "Send the enterprise case study to the prospect." The task appears in your StageWhisper session panel for approval. Once you approve it, it's sent to your AI assistant for execution.

After the call, the assistant can handle follow-ups based on the conversation: drafting recap emails, creating CRM entries, scheduling next meetings, or anything else the assistant is configured to do.

## How it works

<Steps>
  <Step title="Pair your AI assistant">
    In StageWhisper settings, go to the Assistant tab and start the pairing process. You'll get a short pairing code that you enter in your AI assistant's dashboard.
  </Step>

  <Step title="Tasks are detected during calls">
    While you're in a listening session, the AI analyzes the conversation and identifies moments where a task could be created.
  </Step>

  <Step title="You approve or dismiss">
    Each detected task appears in the Assistant tab of your session panel. You can approve it (sends to your assistant), dismiss it, or ignore it.
  </Step>

  <Step title="Your assistant executes">
    Approved tasks are relayed to your connected AI assistant, which carries them out based on its own capabilities and configuration.
  </Step>
</Steps>

## What data is shared

What your assistant sees depends on your reasoning mode:

**BYO mode:** All content is end-to-end encrypted between your desktop and your AI assistant plugin. The backend relays encrypted envelopes and only sees routing metadata (task status, session ID, timestamps).

**Hosted mode:** When a task is relayed to your assistant, StageWhisper shares the transcript excerpt relevant to the task, a signal summary explaining why the task was created, your active Playbook context, and a tone snapshot (if available).

Raw audio is never shared in either mode. See [privacy](/guides/privacy) for the full breakdown.

## Approval modes

By default, every task requires your explicit approval before it's sent to the assistant. You can also configure auto-dispatch for specific task types you trust, so they go straight through without interrupting your call.

Read more in [task automation](/agents/task-automation).

## Getting started

<Columns cols={2}>
  <Card title="Connect OpenClaw" icon="plug" href="/agents/openclaw">
    Pair your OpenClaw host with StageWhisper.
  </Card>

  <Card title="BYO Protocol" icon="code" href="/agents/byo-protocol">
    Implement the reasoning protocol on any AI assistant.
  </Card>

  <Card title="Task automation" icon="list-check" href="/agents/task-automation">
    Configure how tasks are detected, approved, and dispatched.
  </Card>

  <Card title="Run a local model" icon="microchip" href="/agents/local-model">
    Skip the external assistant and run a model on your own device.
  </Card>
</Columns>