***

title: Create with AI
sidebarTitle: Create with AI
description: 'Describe what you need, and AI generates a complete agent for you.'
---------------------------------------------------------------------------------

Don't want to start from scratch? Describe what you want in four prompts, and our AI will generate a complete Single Prompt agent — ready for you to customize.

***

## Getting There

<Steps>
  <Step title="Open Create Agent">
    From your dashboard, click the green **Create Agent** button in the top right.
  </Step>

  <Step title="Choose Create with AI">
    Select the third option in the modal.

    <Frame caption="Creation method selection">
      <img src="https://files.buildwithfern.com/smallest-ai.docs.buildwithfern.com/d435ccacade2625813dd4b2cd3ed434a48c21d9be1eec02ae4ad104de5dca4a9/products/atoms/pages/platform/building-agents/images/create-agent-modal.png" alt="Create agent modal" />
    </Frame>
  </Step>

  <Step title="Select Single Prompt">
    Choose **Single Prompt** as your agent type.

    <Frame caption="Create with AI interface">
      <img src="https://files.buildwithfern.com/smallest-ai.docs.buildwithfern.com/a316f92325c1cd0738efdb710d62ee6ea8bbf940f557e72eb34c337c21e5dbf2/products/atoms/pages/platform/building-agents/images/create-with-ai.png" alt="Create with AI page" />
    </Frame>
  </Step>
</Steps>

***

## Configure Agent (Left Panel)

Set up the basics before writing your prompts:

| Field              | What to Choose                            |
| ------------------ | ----------------------------------------- |
| **Agent Type**     | Single Prompt (already selected)          |
| **Call Direction** | Inbound or Outbound                       |
| **Emotive Model**  | Toggle on for emotional expression (Beta) |
| **Voice**          | Pick from the voice library               |
| **Knowledge Base** | Optionally attach an existing KB          |

***

## The Four Prompts (Right Panel)

Each prompt helps AI understand a different aspect of your agent. All require at least 50 characters.

<Accordion title="1. Role & Objective">
  *Who is this agent and what's their goal?*

  **Tips:**

  * Be specific about the company/industry
  * Define the primary goal clearly
  * Include personality traits (friendly, professional, empathetic)
  * Mention what success looks like
</Accordion>

<Accordion title="2. Conversational Flow">
  *What steps should the agent follow?*

  **Tips:**

  * Think about the typical call from start to finish
  * Include verification steps if needed
  * Mention key decision points (what changes the path?)
  * End with how to close the conversation
</Accordion>

<Accordion title="3. Dos, Don'ts & Fallbacks">
  *How should the agent behave, especially in tricky situations?*

  **Tips:**

  * List explicit "always do" behaviors
  * List explicit "never do" constraints
  * Think about edge cases: what if verification fails? caller gets frustrated? agent doesn't know the answer?
  * Define escalation paths
</Accordion>

<Accordion title="4. End Conditions">
  *When should the call end?*

  **Tips:**

  * Issue resolved and confirmed
  * Customer says goodbye or thanks
  * Successful transfer to human
  * No response after multiple attempts
</Accordion>

***

## Pre-Built Prompts

Not sure what to write? Click any template tab to pre-fill all four prompts:

* Real Estate
* Credit Card
* Customer Support Electronics
* General Sales
* Onboarding Assist

Use **Clear All** to reset and start fresh.

***

## After You Click Create

AI generates your agent and opens the editor with everything already set up:

* **Prompt** — fully written based on your four inputs
* **Voice** — the one you selected
* **Model** — configured and ready
* **Knowledge Base** — attached if you chose one

From there, refine as needed and test.

<CardGroup cols={2}>
  <Card title="Prompt Editor" icon="pen" href="/atoms/atoms-platform/single-prompt-agents/prompt-section/writing-prompts">
    Review and refine the generated prompt
  </Card>

  <Card title="Test Your Agent" icon="flask" href="/atoms/atoms-platform/analytics-and-logs/testing">
    See how it performs
  </Card>
</CardGroup>