AI Personas
The AI profile the user will be roleplaying with
There are two main parts of each AI bot: the persona and the background.
AI Persona
Overview: The AI’s name, appearance, voice, and personality
- Name - The AI’s name
- Appearance - Either the AI’s profile picture or live avatar option
- Voice - The voice model the AI uses (see the drop down menu at the top of each provider’s tab to see price estimates)
- Personality - The AI’s personality type, style of speaking, confidence, professionalism, assertiveness, etc.
Do not include any background or situational context in the Persona. That belongs in the Background.
AI Background
Definition: A general field that lets you input the AI’s relevant situation, background, pains, needs, etc.
Components: There are only two fields for the AI background:
- Title - The AI’s role and company
- Situation/context - Any relevant context to the situation. Common data included here for sales scenarios is:
- Pain points/Specific needs
- Budget/income level
- Decision criteria
- Decision makers
- Competition knowledge
🛠️ Configuration Options Overview
Replay offers three core configuration options for delivering agent experiences. Each option serves different goals depending on user preference, performance needs, and visual requirements.
| Configuration | Best For | Key Features |
|---|---|---|
| Default Voice | When you need a flexible, high-quality voice experience with screen sharing but no video. |
|
| Real-Time API | Live, real-time conversations where speed and responsiveness are critical. |
|
| Video Avatar | Engaging users with a visual, lip-synced avatar for a more personal touch. |
|
🔊 1. Default Voice Option
Overview: This is the most flexible option for users who want to deliver a high-quality voice experience without video.
Key Features:
- Provider Choice: Users can select their preferred TTS provider based on quality, latency, and price.
- Customizable: Supports a wide range of providers, from fast/premium to slower/more cost-effective.
- Screen Sharing: Fully compatible with screen sharing for guided walkthroughs and demos.
- No Video: Does not support real-time webcam video, keeping bandwidth and complexity low.
Best For: When users need spoken responses paired with screen interaction, without the need for a visual avatar or webcam feed.
⚡ 2. Real-Time API Option
Overview: A low-latency, high-speed configuration for real-time experiences that require responsiveness and visual interaction.
Key Features:
- Instant Response: Prioritizes speed for seamless, natural conversations.
- Live Video Feed Support: Enables agents to view the user’s screen or webcam in real-time, if permissions are granted.
- Fixed Provider: Not customizable like the default option—comes with a fixed voice/interaction pipeline.
Best For: Live support scenarios, debugging, or any real-time collaborative sessions where fast back-and-forth is critical.
🧑💻 3. Video Avatar Option
Overview: This configuration introduces a visually engaging avatar that lip-syncs with the agent’s speech.
Key Features:
- Lip-Synced Visuals: Adds an animated, talking avatar to make the experience more immersive and personal.
- Stackable: Can be added on top of either the Default Voice or Real-Time API configurations.
- Customizable Appearance: Users can often customize the look and style of the avatar to match brand or tone.
Best For: Use cases where visual presence increases engagement—like product walkthroughs, onboarding, or virtual assistants.
Creating Variety
You can mix and match personas and backgrounds to create diverse training roleplays:
Different Difficulty Levels
- Easy: “Always say yes”
- Medium: “Quiet, slightly stubborn, but able to be convinced”
- Hard: “Be rude, interrupt the user, etc.”
Different Buyer Types
- Technical buyer: Focuses on specifications and integration
- Financial buyer: Focuses on ROI and cost savings
- End-user buyer: Focuses on ease of use and features