> ## 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.
# Roleplay Training
> Practice sales calls with AI-played HCP personas and get an instant, AI-graded scorecard.
Roleplay Training is the RoseRx Platform's practice simulator for pharmaceutical field teams. You make a simulated voice call to an AI-played healthcare professional (HCP) persona, pitch your product, and receive an instant scorecard — dimension-by-dimension scores, coaching notes, and a full transcript — so you can sharpen your call craft before it counts.
Navigate to **Sales Engagement** → **Training** in the left sidebar to open the **Roleplay Training** page. Sales Engagement is an optional module — contact your RoseRx representative to enable it for your organization.
Roleplay Training is a practice simulator. It is separate from the [Field Team Co-Pilot](/guides/agents/creating-an-agent) agent type, which supports reps live during real HCP conversations. Practice calls are voice-based, conducted in English, and require microphone access in your browser.
## How it works
Pick the product you want to pitch, the HCP persona you want to face, and how long the call should run.
Speak with the persona in real time, just like a phone call. The persona stays in character, raises objections, and paces itself to your chosen call time.
When the call ends, the platform retrieves the transcript and scores your performance against your organization's products, compliance rules, and call methodology.
See your overall score, a breakdown across seven dimensions, coaching notes, and the full transcript.
## Starting a practice call
Click the **Practice** button in the page header to open the **Start a Practice Call** screen, then make three selections:
1. **Product** — the product card you'll pitch. Product cards carry the approved indications, key messages, and claims your call is scored against.
2. **HCP Persona** — the AI-played healthcare professional you'll speak with, listed by name and specialty. Selecting a persona shows a preview card with their specialty and personality.
3. **Call time** — how long the call runs.
| Call time | Length |
| ---------------- | ------- |
| Very short | 2 mins |
| Short | 5 mins |
| Medium (default) | 10 mins |
| Long | 20 mins |
| Very long | 30 mins |
Click **Start Call** to begin. The button stays disabled until you've selected both a product and a persona.
If no options appear, your organization hasn't finished setup yet: "Active personas and product cards are required before calls can start." An admin needs to complete [Setting Up Training](/guides/sales-engagement/setting-up-training) first.
## During the call
The call runs in a full-screen, video-call-style view with the persona's avatar, an elapsed timer, and your chosen call time. While connected you can:
* **Mute / unmute** your microphone
* **End call** when you're done
* **Show or hide the transcript** — a live panel labeling each turn **Persona** or **Rep**
The persona answers with a neutral greeting — "Hello, \[name] speaking." — and waits for you to introduce yourself, just like a real cold call. From there it stays in character, pushes back based on its experience and agreeability settings, and naturally wraps up as your chosen call time runs out.
Treat the opening like a real call: introduce yourself, state your purpose, and ask permission before launching into your pitch. Opening quality is one of the seven scored dimensions.
## After the call
When you end the call, the platform fetches the transcript and scores it automatically — you'll see "Scoring your call…" followed by a "Scorecard ready" confirmation. If a step fails (for example, a dropped connection), the screen offers retry buttons so you can recover without losing the session.
## Your scorecard
The scorecard shows a large **Overall score** out of 100, along with the persona, product, and call time for the session and a **Short**, **Medium**, or **Long** call badge.
### Dimension breakdown
Each session is rated on seven dimensions, each scored 0–100 and shown as a progress bar:
| Dimension | What it measures |
| --------------------- | ------------------------------------------------------------ |
| Product knowledge | Command of the product's data, indications, and key messages |
| Competitor comparison | How well you positioned the product against alternatives |
| Clinical accuracy | Accuracy and compliance of the clinical claims you made |
| Opening quality | Your introduction, stated purpose, and permission to proceed |
| Probing | Questions that uncovered the HCP's experience and barriers |
| Objection handling | How you acknowledged and resolved the persona's objections |
| Commitment | Whether you closed with a concrete next step |
### How the overall score is calculated
The overall score is a weighted average of the seven dimensions. Weights come from the product card you selected; if the card has no custom weights, sensible defaults apply based on your call length — shorter calls emphasize your opening and close, longer calls emphasize product knowledge and objection handling.
Clinical accuracy caps your maximum overall score. Off-label statements, prohibited phrases, or inaccurate safety information limit your final score no matter how strong the rest of the call was — compliance comes first, exactly as it does in the field.
Calls that are too short to assess receive minimum scores with a coaching note asking you to complete a longer roleplay before requesting a score.
### Coaching notes and transcript
Below the dimension breakdown, the scorecard lists up to three actionable **Coaching notes** — specific things to do differently next time — and a **Transcript reference** of the full conversation. Click **New call** to go again.
## Best practices
Rotate across specialties, experience tiers, and agreeability levels. A call that works on an agreeable early-career GP won't survive a hostile KOL.
Start with short calls to drill your opening and close, then move to longer calls that test probing, data depth, and objection handling.
Personas raise the objections your admins configured from real field experience. Rehearse until your approved responses come naturally.
Review your score trend and session history in [Progress & Leaderboard](/guides/sales-engagement/progress-and-leaderboard) to see which dimensions are improving.