Assistants

Main Interface

Create and manage multiple AI assistants, each with their own unique personality, voice, and behavior. Configure every aspect of your assistant through an intuitive interface divided into specialized configuration tabs.

Configuration Tabs

Initial Configuration

string

The opening message your assistant will use. You can enter text for a spoken message or provide an audio file URL (supports mp3, wav formats) for a pre-recorded greeting.

select

Choose how your assistant starts conversations:

Assistant Waits for User: Remains silent until the user speaks first
Assistant Speaks First: Begins with the specified first message
Assistant Speaks First (Generated): Creates a contextual opening message automatically

Context & Knowledge

textarea

Define your assistant’s core behavior and personality. These instructions shape how the assistant understands and responds throughout conversations. Think of it as setting the ground rules for your assistant’s behavior.

select

Select documents your assistant can reference during conversations. This helps your assistant provide accurate, context-aware responses based on your specific content. Manage your documents in the Knowledge Base section.

Response Configuration

slider

default:"0.7"

Controls how creative your assistant’s responses are:

Lower (0.0-0.3): More focused, consistent responses
Medium (0.4-0.7): Balanced creativity
Higher (0.8-1.0): More varied, creative responses

number

default:"250"

Limits the length of your assistant’s responses. A higher number allows for longer, more detailed responses, while a lower number keeps responses concise. One token is approximately 4 characters.

Advanced Options

switch

Improves response speed by remembering answers to similar questions. When enabled, your assistant can quickly provide consistent answers to frequently asked questions.

switch

Allows your assistant to recognize user emotions (like happiness, frustration, or confusion) and adjust responses accordingly for more empathetic interactions.

Basic Voice Settings

select

Choose the voice that best represents your assistant. Each voice has its own unique characteristics and personality, helping create a distinct identity for your assistant.

select

Add ambient sound to make conversations feel more natural:

Office: Creates a professional office environment (default for phone calls)
Off: No background sound (default for web calls)

Speech Processing

number

Sets the minimum text length before converting to speech. A smaller number means faster responses but potentially more choppy speech, while a larger number means smoother speech but slightly delayed responses.

multiselect

Choose which punctuation marks will break speech into natural chunks:

Period (.): Complete thought breaks
Comma (,): Brief pauses
Exclamation (!): Emotional breaks
Question Mark (?): Question breaks
Colon (:): Introductory breaks
Right Parenthesis (): Aside breaks
Hyphen (-): Soft breaks
Vertical Bar (|): Custom breaks
Double Vertical Bar (||): Strong custom breaks

Natural Conversation Features

switch

Adds natural speech fillers (like “um”, “uh”) to make the assistant sound more human-like. This helps create more natural-sounding conversations by mimicking real speech patterns.

switch

Enables subtle listener responses (like “mhmm”, “yeah”) while the user is speaking. This creates more engaging conversations by showing active listening, just like a real person would.

Voice Quality Settings

switch

Enhances voice consistency and character at a slight speed cost. Enable this when voice quality is more important than generation speed.

slider

Controls voice consistency:

Lower (0.0-0.3): More natural, varied speech
Medium (0.4-0.7): Balanced consistency
Higher (0.8-1.0): Very consistent but might sound monotone Tip: Use lower values for longer conversations to sound more natural.

Advanced Voice Settings

slider

Balances speech clarity with voice matching:

Lower: Prioritizes natural speech flow
Higher: Focuses on matching the selected voice Find the sweet spot that sounds best for your use case.

slider

Adjusts voice personality strength:

Lower: More neutral, faster generation
Higher: Stronger personality, more expressive Start low and increase if you want more character in the voice.

Performance Optimization

slider

Balance between response speed and voice quality:

Level 1: Highest quality, longer delays
Level 2-3: Balanced performance
Level 4: Fastest responses, might affect quality Choose higher levels for real-time conversations, lower for pre-recorded messages.

Privacy Settings

switch

Enables healthcare privacy compliance mode:

No storage of logs or recordings
Automatic disabling of audio recording
Organization-wide enforcement Essential for healthcare-related applications.

switch

Store conversation recordings for:

Quality assurance
Training purposes
Compliance requirements Note: Automatically disabled when HIPAA Compliance is on.

Speaking Behavior

slider

Fine-tune how long your assistant pauses before responding:

0 seconds: Immediate responses
1 second: Natural conversation pace
2 seconds: More contemplative style Tip: Shorter times feel more responsive, longer times feel more natural.

switch

Enable AI-powered conversation timing:

Automatically detects when to respond
Adapts to user’s speaking style
Handles pauses intelligently Recommended for most use cases.

Note: When enabled, the manual timing controls below will be hidden as they are automatically managed.

slider

Wait time after complete sentences:

0 sec: Quick responses
1-2 sec: Natural pauses
3 sec: Extended reflection Note: Only available when Smart Endpointing is disabled.

slider

Wait time after incomplete thoughts:

0-1 sec: Quick interjections
1.5 sec: Default pause
2-3 sec: Patient waiting Note: Only available when Smart Endpointing is disabled.

slider

Wait time after numerical input:

0-0.3 sec: Quick confirmation
0.5 sec: Standard pause
1+ sec: Extended verification Note: Only available when Smart Endpointing is disabled.

Interruption Handling

slider

Words needed to interrupt assistant:

0: Any sound interrupts
3-5: Short phrases
7-10: Complete thoughts

slider

Speaking duration to trigger interruption:

0 sec: Instant response
0.2-0.3 sec: Quick acknowledgment
0.5 sec: Confirmed speech

slider

Pause after being interrupted:

0-2 sec: Quick recovery
3-5 sec: Polite pause
6-10 sec: Extended waiting

Call Duration Controls

slider

Automatic call end after silence:

10-30 sec: Quick timeout
60-120 sec: Standard wait
300-600 sec: Extended patience Best Practice: Set based on your use case.

slider

Longest allowed call duration:

300 sec (5 min): Quick interactions
1800 sec (30 min): Standard calls
43200 sec (12 hrs): Extended sessions Tip: Set based on typical conversation length.

Server Integration

url

Endpoint for receiving server messages:

Must be a secure HTTPS endpoint
Handles real-time updates
Processes call events

multiselect

Choose which events to send to your server:

End of Call Report: Final call summary
Function Call: Tool usage events
Conversation Update: Dialog progress
Model Output: AI responses
Speech/Transcript: Voice-to-text
Tool Calls: Integration activities Select only the events you need to process.

Call Messages

string

Custom message when detecting voicemail: Example: “I notice this is a voicemail. I’ll end the call now and try again later.”

string

Polite message before ending calls: Example: “Thank you for your time. Have a great day!”

Idle Handling

multiselect

Select messages to use during silence:

Engagement prompts
Helpful suggestions
Friendly reminders Tip: Choose messages matching your assistant’s personality.

slider

Limit repeated prompts:

1-2: Minimal prompting
3-5: Standard engagement
6-10: Persistent assistance

slider

Seconds before showing idle message:

5-6 sec: Quick engagement
7-8 sec: Natural timing
9-10 sec: Patient waiting

Call Summary

textarea

Configure how your assistant summarizes each call. The summary appears in call.analysis.summary.Sample Format:

Summarize the key points, decisions, and action items from the call in 2-3 sentences.
Focus on customer requirements, solutions provided, and next steps agreed upon.

Tips for effective summaries:

Keep instructions clear and concise
Focus on key information
Specify desired length and format
Include important details to capture

slider

Maximum time allowed for summary generation:

Default: 10 seconds
Range: 1-20 seconds

Note: Request will timeout if processing exceeds this duration.

Success Evaluation

textarea

Define how your assistant evaluates call success. Customize the prompt to match your quality criteria.Recommended Structure:

Analyze the call for:
Customer satisfaction level
Problem resolution completeness
Protocol adherence and professionalism

Best practices:

Define clear success criteria
Include specific metrics
Consider customer satisfaction
Account for business goals

select

Choose how to measure call success:Available Scales:

Numeric Scale: Rate calls from 1-10 Example: “7/10 - Good customer service”
Descriptive Scale: Excellent → Poor Example: “Good - Resolved issue efficiently”
Checklist: Yes/No criteria list Example: “Greeting ✓, Solution ✓, Follow-up ✗”
Matrix: Multi-criteria evaluation Example: “Service: High, Knowledge: Medium”
Percentage Scale: 0-100% Example: “85% successful interaction”
Likert Scale: Strongly Agree → Disagree Example: “Agree - Met customer needs”
Automatic Rubric: AI-generated criteria Example: “Communication: 8/10, Resolution: 9/10”
Pass/Fail: Simple yes/no outcome Example: “true - Call objectives met”

slider

Maximum time allowed for success evaluation:

Default: 10 seconds
Range: 1-20 seconds

Note: Evaluation will stop if processing exceeds this duration.

Structured Data Extraction

textarea

Configure how to extract specific data points from calls.Prompt Guidelines:

Extract the following information:
- Customer contact details
- Product specifications discussed
- Service requirements identified
- Action items and deadlines

Tips for data extraction:

List specific fields to extract
Define expected formats
Include validation rules
Specify required vs optional data

slider

Maximum time allowed for data extraction:

Default: 10 seconds
Range: 1-20 seconds

Note: Extraction will terminate if processing exceeds this duration.

Schema Properties

custom

Define the structure of data to extract from calls.Property Configuration:

Click “Add Property” for each data point
Fill in the details:
- Name: Unique identifier (e.g., “customerName”)
- Type: Data format
  - string: Text (names, descriptions)
  - number: Numerical values
  - boolean: True/False values
  - array: Lists of items
  - object: Nested data structures
- Required: Check if mandatory
- Description: Explain the property

Example Schema:

{
  "customerName": {
    "type": "string",
    "required": true,
    "description": "Full name of the customer"
  },
  "satisfactionScore": {
    "type": "number",
    "required": false,
    "description": "Rating from 1-10"
  }
}

Best Practices:

Use clear, descriptive names
Add helpful descriptions
Mark essential fields as required
Choose appropriate data types

Changes to assistant configurations take effect immediately for new conversations.

Get Started

Guides

Quick Start

Main Interface

Configuration Tabs

Initial Configuration

Context & Knowledge

Response Configuration

Advanced Options

Basic Voice Settings

Speech Processing

Natural Conversation Features

Voice Quality Settings

Advanced Voice Settings

Performance Optimization

Basic Settings

Keyword Recognition

Intelligent Features

Tool Integration

Call Control

Call Routing

Call Termination

Privacy Settings

Speaking Behavior

Interruption Handling

Call Duration Controls

Server Integration

Call Messages

Idle Handling

Call Summary

Success Evaluation

Structured Data Extraction

Schema Properties

​Main Interface

​Configuration Tabs

​Initial Configuration

​Context & Knowledge

​Response Configuration

​Advanced Options

​Basic Voice Settings

​Speech Processing

​Natural Conversation Features

​Voice Quality Settings

​Advanced Voice Settings

​Performance Optimization

​Basic Settings

​Keyword Recognition

​Intelligent Features

​Tool Integration

​Call Control

​Call Routing

​Call Termination

​Privacy Settings

​Speaking Behavior

​Interruption Handling

​Call Duration Controls

​Server Integration

​Call Messages

​Idle Handling

​Call Summary

​Success Evaluation

​Structured Data Extraction

​Schema Properties

Main Interface

Configuration Tabs

Initial Configuration

Context & Knowledge

Response Configuration

Advanced Options

Basic Voice Settings

Speech Processing

Natural Conversation Features

Voice Quality Settings

Advanced Voice Settings

Performance Optimization

Basic Settings

Keyword Recognition

Intelligent Features

Tool Integration

Call Control

Call Routing

Call Termination

Privacy Settings

Speaking Behavior

Interruption Handling

Call Duration Controls

Server Integration

Call Messages

Idle Handling

Call Summary

Success Evaluation

Structured Data Extraction

Schema Properties