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

# AI Agent and AI Outbound

## Configure AI Agent and Run AI Outbound

<Frame>
  <img src="https://mintcdn.com/solvea-a5b10e8e/EDEpJCc6-6I7r3hB/images/image-194.png?fit=max&auto=format&n=EDEpJCc6-6I7r3hB&q=85&s=889efb46c889142320d6ce19e66f606f" alt="Image" width="1600" height="900" data-path="images/image-194.png" />
</Frame>

Configure the AI agent first, test it, then create outbound batches with a focused prompt and recipient list.

### Configure an AI Agent

1. Open **Settings > Agent**.
2. Select the agent you want to edit.
3. Update the AI identity, prompt, and voice.
4. Choose answer behavior for voice agents.
5. Save changes.
6. Test with WebCall before using the agent live.

### Good AI Prompt Logic

A prompt should tell the AI:

* Who it represents.
* What goal it should complete.
* What information it should collect.
* What it should not promise.
* When it should hand off or end the call.

### Create an AI Outbound Batch

<Frame>
  <img src="https://mintcdn.com/solvea-a5b10e8e/EDEpJCc6-6I7r3hB/images/image-195.png?fit=max&auto=format&n=EDEpJCc6-6I7r3hB&q=85&s=ffcafc986799b4fa59ee1023415fa1af" alt="Image" width="1600" height="1000" data-path="images/image-195.png" />
</Frame>

1. Open **Outbound**.
2. Create a new batch.
3. Choose the calling number and AI agent.
4. Add recipients manually or by CSV.
5. Write a batch-specific greeting and prompt.
6. Test the prompt.
7. Launch.
8. Review completed, failed, and pending results.

Use outbound batches for one clear goal at a time. A reminder campaign, qualification campaign, and win-back campaign should be separate batches.