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

# Creating an Agent

> Learn how to create and configure AI-powered agents for healthcare professionals, patients, and field teams in the RoseRx Platform.

Agents are the core building blocks of the RoseRx Platform. Each agent is an AI-powered conversational assistant that delivers accurate, compliant information to your target audience. Before you create one, it helps to understand the four types available.

## Agent types

The RoseRx Platform offers four purpose-built agent types, each designed for a specific audience and use case.

<CardGroup cols={2}>
  <Card title="HCP Clinical Decision Support" icon="user-doctor">
    Built for physicians and primary care providers. Delivers jurisdiction-aware clinical information, dosing guidelines, evidence-based recommendations, and formulary context. Uses professional medical terminology and follows standard compliance rules.
  </Card>

  <Card title="Patient Navigator" icon="heart-pulse">
    Built for patients and consumers. Communicates in simple, accessible language that avoids clinical jargon. Follows strict compliance rules and never provides direct medical advice. Designed to inform and support — not diagnose or prescribe.
  </Card>

  <Card title="Medication Navigator" icon="pills">
    Built for patients on a prescribed medication. Helps them understand how to take it, interactions, side effects, and when to seek help — always reinforcing the role of their prescriber and pharmacist. Follows strict compliance rules.
  </Card>

  <Card title="Field Team Co-Pilot" icon="briefcase-medical">
    Built for pharma sales reps mid-conversation with an HCP. Delivers fast, confident, compliance-safe tips drawn from clinical and competitor data — so reps can get unstuck in seconds. Follows strict compliance rules.
  </Card>
</CardGroup>

### Key differences at a glance

| Feature              | HCP Clinical Decision Support       | Patient Navigator           | Medication Navigator        | Field Team Co-Pilot                  |
| -------------------- | ----------------------------------- | --------------------------- | --------------------------- | ------------------------------------ |
| **Audience**         | Physicians & primary care providers | Patients and consumers      | Patients on a medication    | Pharma sales reps                    |
| **Language**         | Professional medical terminology    | Simple, accessible language | Simple, supportive language | Fast, natural talking points         |
| **Compliance level** | Standard                            | Strict                      | Strict                      | Strict                               |
| **Focus field**      | Product                             | Indicated conditions        | Product                     | Product being promoted + competitors |

<Info>
  The agent type you choose determines which configuration options are available throughout the setup process, including the tone options and focus fields on [General Settings](/guides/agents/general-settings).
</Info>

## Create a new agent

From the left sidebar, click **Agents**, then click **Create Agent**. The **New Agent** dialog offers two starting points:

<Tabs>
  <Tab title="Create from Product Information">
    The fastest way to get a working agent. Link a product and the platform builds the agent for you:

    <Steps>
      <Step title="Link a product">
        * **United States organizations** — Search the **DailyMed** registry by brand or generic name (e.g. "Ozempic") and select your product.
        * **Australian organizations** — Upload the **Product Information PDF** (PDF format, up to 50 MB).
      </Step>

      <Step title="Analysis">
        The platform analyzes the product information with AI — this typically takes 10–20 seconds.
      </Step>

      <Step title="Review the extracted Product Information">
        Check the extracted details before continuing.
      </Step>

      <Step title="Choose the audience and region">
        Select whether the agent serves **HCPs** or **Patients**, and confirm the region (**Australia** or **United States**). The platform checks that the product's jurisdiction matches your organization's region.
      </Step>

      <Step title="Agent created">
        The agent is created with its settings pre-filled and the product information indexed as knowledge sources — ready to test in the Playground.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Blank Agent">
    Start from scratch. A new agent is created immediately in **Draft** status with an auto-generated name, and you're taken to the agent builder to configure everything yourself — starting with [General Settings](/guides/agents/general-settings).
  </Tab>
</Tabs>

<Note>
  The Product Information wizard creates **HCP Clinical Decision Support** or **Patient Navigator** agents (via its HCP/Patient audience choice). To set up a **Medication Navigator** or **Field Team Co-Pilot**, select that type under **Agent instructions** in [General Settings](/guides/agents/general-settings) — typically starting from a Blank Agent.
</Note>

<Tip>
  Agents are always created in **Draft** status. Take the time to configure all settings — general, knowledge base, guidance, and more — before publishing your agent.
</Tip>

## Agent statuses

Every agent in the RoseRx Platform has a status that reflects where it is in its lifecycle.

| Status       | Description                                                                                                                                                    |
| ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Draft**    | Newly created. The agent is being configured and is not yet available to end users. You can freely edit all settings.                                          |
| **Active**   | Published and live. The agent is available to end users through any connected widgets. Changes require re-publishing.                                          |
| **Inactive** | Temporarily disabled. The agent is not available to end users but retains all its configuration. Can be reactivated at any time.                               |
| **Archived** | Permanently retired. The agent is removed from active use and cannot be connected to widgets. Archiving is available from the Danger Zone in General Settings. |

<Warning>
  Archiving an agent is a significant action. Archived agents are removed from all connected widgets and can no longer serve responses. Make sure this is what you intend before proceeding.
</Warning>

## Next steps

Once your agent is created, work through each configuration section:

<CardGroup cols={2}>
  <Card title="General Settings" icon="gear" href="/guides/agents/general-settings">
    Name your agent, choose an AI model, set the tone and response length.
  </Card>

  <Card title="Knowledge Base" icon="book" href="/guides/agents/knowledge-base">
    Add documents, URLs, Q\&A pairs, and text snippets to power your agent's responses.
  </Card>

  <Card title="Guidance & Guardrails" icon="shield-check" href="/guides/agents/guidance-and-guardrails">
    Configure regulatory compliance, prohibited terms, and off-label policies.
  </Card>

  <Card title="Adverse Event Monitoring" icon="triangle-exclamation" href="/guides/agents/adverse-event-monitoring">
    Enable AI-powered detection of potential adverse drug reactions.
  </Card>
</CardGroup>