Skip to main content
Version: Next

Key Terminologies

Inbound Call​

Voicing AI can receive customer phone calls. When a call comes in, your pre-configured Voice Agent answers, follows your scripted instructions, and engages in meaningful conversation.

Technical Details:

  • Inbound calls are handled through Deployments that connect phone numbers to AI Assistants
  • Each inbound call is tracked as a Call record with status, duration, transcripts, and analytics
  • Phone numbers are managed through Inbound Phone Numbers and can be assigned to specific deployments
  • Supports multiple telephony providers (Twilio, Plivo, etc.) for call routing

AI Assistant​

An AI Assistant (also called "Voice Agent" or "Agent") is the core conversational AI entity that handles customer interactions. It contains all the configuration needed to conduct intelligent voice conversations.

Key Components:

  • Basic Settings: Name, description, language, model selection, conversation style, and custom greetings
  • TTS Settings: Voice configuration, speech enhancements, punctuation guides, and multi-language support
  • STT Settings: Speech recognition models, voice activity detection, and audio enhancements
  • Orchestration Settings: User idle detection, call transfer logic, call end conditions, and pathway transfers
  • Telephony Settings: Provider configuration, phone number assignment, call recording, and SIP transfer
  • Tools & AI Settings: LLM configuration, custom tools integration, knowledge base connections, and tool execution settings
  • Post-Call Settings: Analysis prompts, webhook notifications, custom metrics, and data extraction

Assistant Modes:

  • Prompt Mode: Uses a single, comprehensive prompt to guide conversations
  • Pathways Mode: Uses structured conversational pathways (flowcharts) to guide conversations step-by-step

Status States:

  • draft: Assistant is being configured
  • live: Assistant is ready but not deployed
  • deployed: Assistant is deployed to a phone number
  • active: Assistant is actively handling calls

Pathways (Conversational Pathways)​

Pathways are structured, intelligent flowcharts that guide your Voice Agent through conversations step-by-step. They provide a visual, node-based approach to designing conversation flows.

Key Components:

  • Nodes: Individual steps in the conversation flow (11 node types available)
  • Edges: Connections between nodes that define conversation paths based on user responses
  • Global Prompt: Overall instructions that apply to the entire pathway
  • Global Variables: Variables that can be extracted and used across the pathway
  • Tools: Custom tools that can be invoked at specific nodes
  • Knowledge Bases: Knowledge bases that can be queried during the conversation

Node Types:

  1. Start Call: Entry point of the conversation (greeting)
  2. Default: Standard conversation step (agent speaks/asks)
  3. Wait for Response: Explicit pause to collect user input
  4. Knowledge Base: Query knowledge base for answers
  5. Webhook: Call external API to fetch/update data
  6. Custom Tool: Execute a configured tool/integration
  7. Transfer Call: Escalate to human agent or external number
  8. Transfer Pathway: Jump to another pathway (reusable subflows)
  9. SMS: Send text message during conversation
  10. Amazon Connect: Integrate with Amazon Connect contact center
  11. End Call: Final step to gracefully end the conversation

Pathway Features:

  • AI Generation: Automatically generate pathways from natural language descriptions using OpenAI or Gemini
  • Version Control: Track pathway versions for iteration and rollback
  • Access Control: Private (user-only) or Team (shared within organization)
  • Status Management: Draft (in progress) or Complete (ready to use)

Usage:

  • Pathways replace traditional single-prompt assistants when you need structured, multi-step conversations
  • Each node can have its own prompt, conditions, and actions
  • Edges define how conversations flow based on user responses (e.g., "not interested" → different path)

Conversational Flow​

Conversational flows are structured, intelligent pathways that guide your Voice Agent through a conversation. They allow you to:

  • Define how the agent should respond at each step of the dialogue.
  • Program complex decision-making scenarios based on user inputs.

For example, in a customer support scenario, the agent can begin with a greeting, collect necessary details, and escalate the call to a human representative if required.

Note: "Conversational Flow" is the general concept, while "Pathways" is the specific implementation in Voicing AI platform.

Knowledge Base​

A Knowledge Base is a repository of documents, FAQs, and information that your AI Assistant can query during conversations to provide accurate, context-aware answers.

Key Features:

  • Source Management: Add documents via URLs, file uploads, or direct text input
  • RAG (Retrieval-Augmented Generation): Uses advanced search to find relevant information and generate answers
  • Status Tracking: Monitor indexing status (pending, processing, completed, failed) for each source
  • Statistics: Track document count, total tokens, and indexing metrics
  • Access Control: Private (user-only) or Team (shared within organization)

How It Works:

  1. Upload documents or add URLs to your knowledge base
  2. System indexes the content for fast retrieval
  3. During conversations, the assistant queries the knowledge base when needed
  4. Relevant information is retrieved and used to generate accurate responses

Use Cases:

  • Customer support FAQs
  • Product documentation
  • Company policies and procedures
  • Training materials
  • Any structured information your assistant needs to reference

Custom Tools​

Custom tools let you integrate your business operations directly into calls. You can design tools to send alerts, create support tickets, update systems, and more.

Tool Types:

  • Webhook Tools: HTTP-based integrations with external APIs
  • MCP Tools: Model Context Protocol tools for advanced integrations
  • Function Tools: Custom functions that can be called during conversations

Tool Configuration:

  • Authentication: API keys, OAuth, Bearer tokens, custom auth methods
  • Request Configuration: Headers, parameters, body templates, query strings
  • Response Handling: Output variable extraction, error handling, retry logic
  • Timeout & Retry: Configurable timeouts and retry attempts

Integration Methods:

  • REST API Calls: Connect to any RESTful service
  • MCP Server Integration: Connect to MCP-compatible servers
  • Custom Functions: Define custom logic for specific use cases

Use Cases:

  • Create support tickets in your CRM
  • Check order status in your e-commerce system
  • Send notifications to Slack/Teams
  • Update customer records
  • Trigger workflows in other systems

Webhooks​

Webhooks enable Voice Agents to perform live, external actions by connecting to your APIs. For instance, during a call, the agent can book an appointment or fetch customer data from your CRM.

Webhook Node Features:

  • HTTP Methods: GET, POST, PUT, PATCH, DELETE
  • Request Configuration: Custom headers, body, query parameters
  • Response Extraction: Extract specific data from API responses using JSON paths
  • Conditional Routing: Route conversations based on webhook response data
  • Timeout Control: Configurable timeout values (default: 10 seconds)
  • Variable Extraction: Extract variables from responses for use in conversation

Webhook vs Custom Tools:

  • Webhooks are configured within pathway nodes for specific conversation steps
  • Custom Tools are reusable integrations that can be called from multiple places

Campaigns​

A campaign is a group of calls managed collectively. Upload contact lists in formats like CSV to launch outbound campaigns tailored for sales, surveys, or reminders.

Campaign Features:

  • Contact Management: Link to contact groups for bulk calling
  • Assistant Assignment: Assign specific AI Assistant to handle campaign calls
  • Retry Logic: Configure maximum retries per day and retry intervals
  • Re-run Capability: Schedule automatic re-runs with frequency and time settings
  • Status Tracking: Monitor campaign status (pending, running, completed, paused)
  • Call Analytics: Track call outcomes, sentiment, and extracted data

Campaign Configuration:

  • Telephony Settings: Configure provider and call settings
  • Scheduling: Set live time and re-run schedules
  • Retry Strategy: Define retry attempts and intervals
  • Access Control: Private or Team sharing

Use Cases:

  • Sales outreach campaigns
  • Customer satisfaction surveys
  • Appointment reminders
  • Marketing campaigns
  • Follow-up calls

Deployment​

A Deployment connects an AI Assistant to a phone number, making it available to receive inbound calls. It's the bridge between your assistant configuration and actual phone connectivity.

Deployment Features:

  • Phone Number Assignment: Link assistants to specific phone numbers
  • Operating Hours: Configure when the assistant is available (start/end times, off days)
  • Peak Hours Management: Define peak hours and adjust concurrency accordingly
  • Call Volume Limits: Set monthly call volume limits
  • Concurrency Control: Configure min/max concurrent calls for peak and non-peak hours
  • Uptime Configuration: Set daily and weekly uptime hours

Deployment Status:

  • draft: Deployment is being configured
  • active: Deployment is live and receiving calls
  • paused: Deployment is temporarily disabled
  • archived: Deployment is deactivated

Phone Number Management:

  • Inbound Phone Numbers: Manage phone numbers that can receive calls
  • Provider Support: Works with multiple telephony providers (Twilio, Plivo, etc.)
  • Webhook Configuration: Configure webhooks for inbound call events
  • Number Validation: Validate phone numbers before assignment

Contacts & Contact Groups​

Contacts are individual customer records that can be used in campaigns. Contact Groups are collections of contacts organized for bulk operations.

Contact Features:

  • Contact Information: Name, phone number, email, and custom metadata
  • Grouping: Organize contacts into groups for campaign targeting
  • Bulk Import: Upload contacts via CSV files
  • Custom Fields: Add custom metadata to contacts
  • Access Control: Private or Team sharing

Contact Group Features:

  • Bulk Management: Manage multiple contacts together
  • Campaign Assignment: Assign entire groups to campaigns
  • Filtering: Organize contacts by criteria
  • Import/Export: Bulk import and export capabilities

Organizations​

Organizations enable team collaboration and resource sharing within your Voicing AI workspace.

Organization Features:

  • Team Sharing: Share assistants, pathways, tools, and knowledge bases with team members
  • Role-Based Access: Assign roles (Owner, Admin, Developer, Analyst) with specific permissions
  • Member Management: Invite members, manage invitations, and control access
  • Resource Isolation: Organize resources by organization for better management

Access Types:

  • Private: Resource is only accessible to the creator
  • Team: Resource is shared with all organization members

Member Roles:

  • Owner: Full control over organization and resources
  • Admin: Administrative access to organization settings
  • Developer: Can create and modify resources
  • Analyst: Read-only access for analytics and reporting

Voices​

Voicing AI offers a diverse range of voices to suit different use cases. Choose from conversational, professional, or dynamic tones to best represent your brand during calls.

Voice Features:

  • Voice Selection: Choose from multiple voice providers (ElevenLabs, Voicing, etc.)
  • Voice Customization: Adjust stability, similarity boost, and other parameters
  • Language Support: Voices support multiple languages
  • Speaker Selection: Choose specific speakers/voices for different scenarios
  • Audio Quality: Configure audio quality settings for optimal call experience

Voice Configuration:

  • Voice ID: Unique identifier for each voice
  • Speaker Name: Human-readable name for the voice
  • Language: Primary language of the voice
  • Stability: Control voice consistency (0.0 to 1.0)
  • Similarity Boost: Enhance voice similarity to original (0.0 to 1.0)

TTS (Text-to-Speech)​

TTS converts written text into natural-sounding speech audio that plays during phone calls.

TTS Features:

  • Multiple Providers: Support for ElevenLabs, Voicing, and other TTS providers
  • Voice Models: Choose from various TTS models with different capabilities
  • Audio Generation: Generate audio files from text with customizable settings
  • Record Management: Store and manage generated audio records
  • Pronunciation Guides: Custom pronunciation rules for specific words
  • Speech Enhancements: Speed control, expressiveness, number normalization

TTS Models:

  • Model Selection: Choose from available TTS models based on quality and language
  • Supported Speakers: Each model supports multiple speakers/voices
  • Language Support: Models support various languages
  • Active/Recommended: Models can be marked as active or recommended

TTS Records:

  • Audio Storage: Generated audio files are stored in cloud storage
  • Metadata Tracking: Track text, voice, language, file size, duration
  • Status Monitoring: Monitor generation status (pending, completed, failed)

STT (Speech-to-Text)​

STT converts spoken audio into text transcripts, enabling the AI to understand what callers are saying.

STT Features:

  • Multiple Providers: Support for various STT providers (Google, Azure, Voicing, etc.)
  • Model Selection: Choose from available STT models
  • Language Support: Models support multiple languages
  • Voice Activity Detection: Detect when users are speaking
  • Audio Enhancements: Auto punctuation, profanity filtering, smart formatting

STT Configuration:

  • Recognition Settings: Configure recognition models and supported languages
  • VAD Settings: Voice activity detection thresholds and sensitivity
  • Enhancement Filters: Phrase lists, punctuation, formatting options
  • Interruption Strategy: Configure how the assistant handles user interruptions

STT Models:

  • Model Selection: Choose from available STT models
  • Language Support: Each model supports specific languages
  • Multi-language: Some models support multiple languages simultaneously
  • Active/Recommended: Models can be marked as active or recommended

Telephony​

Telephony refers to the phone system infrastructure that enables voice calls. Voicing AI integrates with multiple telephony providers to handle inbound and outbound calls.

Telephony Providers:

  • Twilio: Popular cloud communications platform
  • Plivo: Cloud telephony API provider
  • Other Providers: Support for additional telephony providers

Telephony Features:

  • Phone Number Management: Acquire and manage phone numbers
  • Call Routing: Route calls to appropriate assistants
  • Call Recording: Record calls for analysis and compliance
  • SIP Transfer: Transfer calls using SIP protocol
  • Webhook Integration: Receive call events via webhooks
  • Provider Configuration: Configure provider-specific settings

Inbound Phone Numbers:

  • Number Assignment: Assign numbers to deployments
  • Provider Management: Manage numbers across different providers
  • Availability Status: Track which numbers are available or assigned
  • Region Support: Support for numbers in different countries/regions

LLM Models​

LLM (Large Language Model) Models are the AI engines that power conversational intelligence in your assistants.

LLM Model Features:

  • Multiple Providers: Support for OpenAI, Gemini, and other providers
  • Model Selection: Choose from various models (GPT-4, Gemini Pro, etc.)
  • Capabilities: Each model has specific capabilities and features
  • Context Windows: Models have different context window sizes
  • Token Limits: Maximum token limits for input and output
  • Costing: Track costs per 1K input/output tokens

LLM Configuration:

  • Model Selection: Select which LLM model to use for your assistant
  • Temperature: Control response creativity (0.0 to 1.0)
  • Max Tokens: Set maximum response length
  • Context Memory: Enable conversation context retention
  • Response Delay: Configure artificial response delays

Model Management:

  • Active Models: Mark models as active for use
  • Recommended Models: Highlight recommended models for users
  • Capability Tracking: Track what each model can do
  • Cost Tracking: Monitor usage costs per model

API Tokens​

API Tokens provide programmatic access to Voicing AI platform APIs for automation and integration.

API Token Features:

  • Token Generation: Create secure API tokens with unique prefixes
  • Token Management: View, regenerate, and revoke tokens
  • Environment Support: Organize tokens by environment (dev, staging, prod)
  • Rate Limiting: Configure rate limits per token
  • Usage Tracking: Monitor API usage and statistics
  • Security: Secure token storage with hashing

Token Lifecycle:

  • Creation: Generate new tokens with descriptions
  • Usage: Use tokens to authenticate API requests
  • Regeneration: Generate new tokens while invalidating old ones
  • Revocation: Revoke tokens when no longer needed
  • Expiration: Set expiration dates for tokens

Calls​

Calls are individual phone call records that track conversations between your AI Assistant and customers.

Call Features:

  • Status Tracking: Track call status (pending, in-progress, completed, failed)
  • Call Metadata: Store call start/end times, duration, provider information
  • Transcripts: Full conversation transcripts with speaker identification
  • Recordings: Links to call recordings for playback
  • Sentiment Analysis: Analyze call sentiment (positive, negative, neutral)
  • Data Extraction: Extract structured data from conversations
  • Retry Logic: Automatic retry for failed calls

Call Analytics:

  • Call Metrics: Duration, status, outcome tracking
  • Sentiment Tracking: Overall call sentiment analysis
  • Variable Extraction: Extract custom variables from conversations
  • Campaign Attribution: Link calls to specific campaigns
  • Performance Metrics: Success rates, average duration, etc.

Analytics​

Analytics provide insights into call performance, assistant effectiveness, and business metrics.

Analytics Features:

  • Call Analytics: Detailed metrics on individual and bulk calls
  • Campaign Analytics: Performance metrics for campaigns
  • Assistant Analytics: Effectiveness metrics for assistants
  • Sentiment Analysis: Track positive/negative/neutral sentiment trends
  • Data Extraction: Analyze extracted variables and data points
  • Export Capabilities: Export analytics data for external analysis

Analytics Metrics:

  • Call Volume: Number of calls made/received
  • Success Rates: Percentage of successful calls
  • Average Duration: Mean call length
  • Sentiment Distribution: Breakdown of call sentiments
  • Variable Completion: Rate of variable extraction success

Sound Mixer​

Sound Mixer allows you to create custom audio backgrounds and sound effects for calls.

Sound Mixer Features:

  • Audio Mixing: Combine multiple audio tracks
  • Background Sounds: Add background music or ambient sounds
  • Sound Effects: Add sound effects at specific moments
  • Audio Editing: Trim, adjust volume, and mix audio tracks
  • Storage: Store mixed audio files for reuse

MCP (Model Context Protocol) Tools​

MCP Tools are advanced integrations using the Model Context Protocol for sophisticated tool interactions.

MCP Features:

  • Server Integration: Connect to MCP-compatible servers
  • Tool Discovery: Automatically discover available tools from MCP servers
  • Protocol Support: Full MCP protocol implementation
  • Transport Types: Support for HTTP and other transport methods
  • Authentication: MCP-specific authentication methods
  • Tool Caching: Cache discovered tools for performance

Assistant Templates​

Assistant Templates are pre-configured assistant configurations that can be used as starting points.

Template Features:

  • Use Case Templates: Templates for common use cases
  • Prompt Templates: Pre-written prompts for specific scenarios
  • Quick Start: Fast assistant creation from templates
  • Customization: Templates can be customized after creation

System Audit Logs​

System Audit Logs track all user actions and system events for security and compliance.

Audit Log Features:

  • Action Tracking: Log all create, update, delete operations
  • User Attribution: Track which user performed each action
  • Resource Tracking: Track which resources were modified
  • Timestamp Tracking: Precise timestamps for all events
  • Compliance: Support for compliance and security requirements

Summary​

Voicing AI platform provides a comprehensive suite of tools for building, deploying, and managing AI-powered voice conversations. From creating assistants and pathways to managing campaigns and analyzing performance, the platform offers everything needed for enterprise-grade voice AI applications.

Core Workflow:

  1. Create an AI Assistant with prompts or pathways
  2. Configure TTS/STT models, tools, and knowledge bases
  3. Deploy the assistant to a phone number via Deployment
  4. Launch campaigns for outbound calls or receive inbound calls
  5. Analyze performance through comprehensive analytics

Key Differentiators:

  • Dual Mode: Choose between prompt-based or pathway-based assistants
  • Flexible Integration: Extensive tool and webhook support
  • Team Collaboration: Organization-based sharing and access control
  • Enterprise Ready: Audit logs, API tokens, and compliance features