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

# Create Agent

> Create an agent (knowledge base or presenter)

## Creating Your First Agent

This guide walks you through the complete process of creating an agent in Aivah's **Agents** workspace (`/agents/create`). Once your agent is created, Aivah automatically trains it on your added content, making it ready for use immediately.

<Frame caption="Create agent form – Name, Agent Capabilities, Agent Prompt, Persona, and Documents">
  <img src="https://mintcdn.com/opinionai/tR4nAAll3jemMkvA/images/agent-create-form.png?fit=max&auto=format&n=tR4nAAll3jemMkvA&q=85&s=9f27e0af48728b797e8e496f9ec1baca" alt="Create agent form" width="4112" height="2658" data-path="images/agent-create-form.png" />
</Frame>

<img className="block dark:hidden" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.13.17%20PM.png" alt="Create New Agent Interface" />

<img className="hidden dark:block" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.13.17%20PM.png" alt="Create New Agent Interface" />

## Prerequisites

Before creating your agent, ensure you have:

* An active Aivah account with **Agents** workspace access
* Your content ready (documents, URLs, text, audio, or video)
* A clear understanding of your agent's purpose

## Step-by-Step Creation Process

### 1. Access the Agents Workspace

<Steps>
  <Step title="Navigate to Agents">
    Sign in to your Aivah account and click **Agents** in the side navigation (`/agents`).
  </Step>

  <Step title="Start Agent Creation">
    Click the **Plus (+)** icon in the agents list header to open `/agents/create`.
  </Step>
</Steps>

### 2. Configure Basic Settings

<Steps>
  <Step title="Name Your Agent">
    Enter a descriptive name in the **Name** field. The name becomes the slug shown next to the heading (for example `@customer-support`).
  </Step>

  <Step title="Pick the Capability Profile">
    Under **Agent capabilities**, decide whether the agent is a **Knowledge base** or a **Presenter**:

    * **Presenter toggle Off** *(default)*: a regular knowledge-base agent. Accepts URLs, documents, text, audio, video, and images.
    * **Presenter toggle On**: a presentation agent. Accepts a single PDF or a single video, produces slide-aligned answers, and supports the in-chat **Quiz** overlay.
  </Step>

  <Step title="Write the Agent Prompt">
    Use **Agent prompt** to give free-form instructions that shape behaviour (for example *"Always answer in Spanish"* or *"Use a friendly tone and short sentences"*).
  </Step>

  <Step title="Set a Persona">
    In **Persona**, give a short label of your agent's tone, such as *witty and casual* or *formal and concise*.
  </Step>
</Steps>

<Warning>
  Choose your agent name carefully – it becomes the slug and is visible to users.
</Warning>

### 3. Add Initial Knowledge

Open the **Add document** dropdown to attach knowledge sources. Knowledge-base agents accept any combination of the methods below; **Presenter** agents accept only a single PDF or video.

<Frame caption="Add document dropdown – Add URL / Add Files / Create Text">
  <img src="https://mintcdn.com/opinionai/tR4nAAll3jemMkvA/images/agent-add-document-dropdown.png?fit=max&auto=format&n=tR4nAAll3jemMkvA&q=85&s=84c356a900553dde5e34d1d908b24231" alt="Add document dropdown" width="4112" height="2658" data-path="images/agent-add-document-dropdown.png" />
</Frame>

<Tabs>
  <Tab title="Add URL">
    **Web Content Integration**

    1. Pick **Add URL** from the **Add document** dropdown.
    2. Paste a publicly accessible link – website pages, blog posts, YouTube/Vimeo videos, and public documentation all work.
    3. Aivah fetches and indexes the page automatically.

    <Frame caption="Add URL modal">
      <img src="https://mintcdn.com/opinionai/tR4nAAll3jemMkvA/images/agent-add-url-modal.png?fit=max&auto=format&n=tR4nAAll3jemMkvA&q=85&s=cba6dfdbba9668ddef50d5bc1fffdbf6" alt="Add URL modal" width="4112" height="2658" data-path="images/agent-add-url-modal.png" />
    </Frame>
  </Tab>

  <Tab title="Add Files">
    **Document Upload**

    1. Pick **Add Files** from the dropdown – drag-and-drop or click to open the file picker.
    2. Supported formats include:
       * **Documents**: `.pdf`, `.docx`, `.txt`, `.csv`, `.xlsx`, `.html`, `.md`, `.epub`, `.json`, `.xml`
       * **Media**: `.mp4`, `.mp3`, `.jpg`, `.png` and more
    3. Each file can be up to **21 MB**.

    <Frame caption="Add Files modal with the full list of supported file types">
      <img src="https://mintcdn.com/opinionai/tR4nAAll3jemMkvA/images/agent-add-files-modal.png?fit=max&auto=format&n=tR4nAAll3jemMkvA&q=85&s=b5d6e570e2e94354f3c10e82995a40b3" alt="Add Files modal" width="4112" height="2658" data-path="images/agent-add-files-modal.png" />
    </Frame>
  </Tab>

  <Tab title="Create Text">
    **Direct Text Entry**

    1. Pick **Create Text** from the dropdown.
    2. Paste raw text content – ideal for FAQs, policies, or short informational snippets.
    3. The text is immediately processed.

    <Frame caption="Create Text modal">
      <img src="https://mintcdn.com/opinionai/tR4nAAll3jemMkvA/images/agent-create-text-modal.png?fit=max&auto=format&n=tR4nAAll3jemMkvA&q=85&s=e828f992a394d5b5327dabdfc421de1b" alt="Create Text modal" width="4112" height="2658" data-path="images/agent-create-text-modal.png" />
    </Frame>
  </Tab>
</Tabs>

### 4. Create and Verify

<Steps>
  <Step title="Click Create agent">
    Press **Create agent**. The upload starts, the agent enters `processing`, and the page redirects back to the agents list.
  </Step>

  <Step title="Watch the live training status">
    The agents list polls every 10 seconds while any agent is in `pending` or `processing` and shows a **Refresh** icon automatically. The status flips to `completed` once training is done.

    <img className="block dark:hidden" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.13.49%20PM.png" alt="Upload Progress Notification" />

    <img className="hidden dark:block" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.13.49%20PM.png" alt="Upload Progress Notification" />
  </Step>

  <Step title="Open the agent detail view">
    Click any row in the agents list to open `/agents/{id}`. From here you can:

    * View every URL, file, or text snippet with its ingestion status (`uploaded`, `processing`, `completed`, `failed`).
    * Retry a single failing source with the inline refresh icon (no need to retrain the whole agent).
    * Open the **Edit** button (top right) to change name, prompt, persona, and add or remove documents.

    <img className="block dark:hidden" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.14.23%20PM.png" alt="Agent Knowledge Settings - List View" />

    <img className="hidden dark:block" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.14.23%20PM.png" alt="Agent Knowledge Settings - List View" />
  </Step>
</Steps>

## Post-Creation Configuration

### Knowledge Management

Once created, you can manage your agent's knowledge:

* **Add Content**: Use the **Add content** button for additional sources
* **Edit Sources**: Click any source to modify or update it
* **Remove Content**: Use trash icons or **Delete All** for bulk removal
* **Monitor Processing**: Ensure all sources show character counts (indicates successful processing)

You can also switch to **Card View** to see a visual representation of your knowledge sources:

<img className="block dark:hidden" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.14.44%20PM.png" alt="Agent Knowledge Settings - Card View" />

<img className="hidden dark:block" src="https://mintlify.s3.us-west-1.amazonaws.com/opinionai/knowledge-agent/Knoweldge%20%20/Screenshot%202025-08-09%20at%201.14.44%20PM.png" alt="Agent Knowledge Settings - Card View" />

### Next Steps

After creating your agent:

1. **Test Interactions**: Verify the agent responds accurately to test questions
2. **Configure Behavior**: Set up response style and limitations in Share settings
3. **Deploy**: Generate sharing links or embeds for your users
4. **Monitor Performance**: Use Insights to track usage and effectiveness

## Best Practices

<AccordionGroup>
  <Accordion title="Content Quality">
    * Use clear, well-structured documents
    * Ensure information is current and accurate
    * Remove outdated or contradictory content
  </Accordion>

  <Accordion title="Knowledge Organization">
    * Group related content together
    * Use descriptive titles for uploaded documents
    * Regularly review and update your knowledge base
  </Accordion>

  <Accordion title="Testing">
    * Test with common user questions
    * Verify accuracy of responses
    * Check for knowledge gaps and fill them
  </Accordion>
</AccordionGroup>

## Common Issues and Solutions

<AccordionGroup>
  <Accordion title="Upload Takes Too Long">
    Large files and some websites require extended processing time. Keep your browser tab open until the upload banner shows completion.
  </Accordion>

  <Accordion title="URL Won't Process">
    Ensure the URL is:

    * Publicly accessible (not behind login)
    * Not geo-restricted
    * Contains readable content (not just images)
  </Accordion>

  <Accordion title="Agent Gives Incorrect Answers">
    * Review your source content for accuracy
    * Remove conflicting information
    * Add more specific examples or context
  </Accordion>
</AccordionGroup>

Ready to enhance your agent further? Learn about [adding more content](/knowledge-agent/add-content) or explore [agent types](/knowledge-agent/types-of-agents) for advanced configurations.