> ## Documentation Index > Fetch the complete documentation index at: https://blackbox.dasha.ai/llms.txt > Use this file to discover all available pages before exploring further. # Run one stateless text-chat completion turn. > Statelessness: the client carries an opaque encrypted `state` blob between turns. The server decodes the state, drives the agent's orchestration loop (OpenAI ChatCompletions + tool calls — built-in, agent-defined webhook, or MCP), and returns the new messages plus a fresh encrypted state. The conversation transcript is owned by the client. ## OpenAPI ````yaml https://blackbox.dasha.ai/swagger/v1/swagger.json post /api/v1/text-chat/completion openapi: 3.0.4 info: title: Dasha BlackBox Agent API description: API for managing AI agents and calls contact: name: DashaAI Team email: support@dasha.ai version: v1 servers: - url: https://blackbox.dasha.ai description: Dasha BlackBox Agent API security: - {} - {} tags: - name: ActivityLogs - name: Agents - name: CallResults - name: Calls - name: Chats - name: CustomerData - name: Mcp - name: Media - name: Misc - name: PronunciationDictionaries - name: Providers - name: SipAliases - name: SipCredentials - name: SipPhoneNumbers - name: TextChat - name: TwilioProvider - name: Voice - name: WebhookTest - name: WebIntegrations - name: WebSocket description: WebSocket endpoints for real-time communication paths: /api/v1/text-chat/completion: post: tags: - TextChat summary: Run one stateless text-chat completion turn. description: >- Statelessness: the client carries an opaque encrypted `state` blob between turns. The server decodes the state, drives the agent's orchestration loop (OpenAI ChatCompletions + tool calls — built-in, agent-defined webhook, or MCP), and returns the new messages plus a fresh encrypted state. The conversation transcript is owned by the client. requestBody: description: >- Completion request with the agent reference, conversation, and prior state. content: application/json: schema: allOf: - $ref: '#/components/schemas/TextChatCompletionRequest' description: >- Stateless text-chat completion request. Carries the full conversation history (client-owned) plus the encrypted server-issued DashaAI.BlackBox.AgentAPI.TextChat.Dtos.TextChatCompletionRequest.State from the previous turn. text/json: schema: allOf: - $ref: '#/components/schemas/TextChatCompletionRequest' description: >- Stateless text-chat completion request. Carries the full conversation history (client-owned) plus the encrypted server-issued DashaAI.BlackBox.AgentAPI.TextChat.Dtos.TextChatCompletionRequest.State from the previous turn. application/*+json: schema: allOf: - $ref: '#/components/schemas/TextChatCompletionRequest' description: >- Stateless text-chat completion request. Carries the full conversation history (client-owned) plus the encrypted server-issued DashaAI.BlackBox.AgentAPI.TextChat.Dtos.TextChatCompletionRequest.State from the previous turn. responses: '200': description: Completion succeeded. content: application/json: schema: $ref: '#/components/schemas/TextChatCompletionResponse' '400': description: >- Request validation failed (bad state, malformed agent reference, etc.). content: application/json: schema: $ref: '#/components/schemas/TextChatErrorDto' '401': description: Missing or invalid API key. content: application/json: schema: $ref: '#/components/schemas/ProblemDetails' '404': description: Agent referenced by id was not found for the organization. content: application/json: schema: $ref: '#/components/schemas/TextChatErrorDto' '502': description: The upstream LLM provider or a tool returned an error. content: application/json: schema: $ref: '#/components/schemas/TextChatErrorDto' components: schemas: TextChatCompletionRequest: required: - agent - messages type: object properties: agent: allOf: - $ref: '#/components/schemas/AgentRefDto' description: Agent to drive the completion — by id or by inline config. messages: type: array items: $ref: '#/components/schemas/ChatMessageDto' description: >- Complete conversation history including the new user message at the end. Tool calls are bundled in the previous assistant turn(s). state: type: string description: >- Opaque encrypted state from the previous turn. Null/omitted on the first turn. nullable: true additional_data: type: object additionalProperties: {} description: >- Free-form variable bag merged into agent prompt template variables (e.g. `{{customerName}}`). Matches the project-wide `AdditionalData` convention. nullable: true finish: type: boolean description: >- Client/user signal to end the conversation. When true, this turn still produces an assistant reply (if any) and then runs the same finish pipeline as the agent's `finishTheConversation` built-in: post-call analysis + completion webhook + persistence + audit events. nullable: true additionalProperties: false description: >- Stateless text-chat completion request. Carries the full conversation history (client-owned) plus the encrypted server-issued DashaAI.BlackBox.AgentAPI.TextChat.Dtos.TextChatCompletionRequest.State from the previous turn. TextChatCompletionResponse: type: object properties: messages: type: array items: $ref: '#/components/schemas/ChatMessageDto' description: >- New messages produced during this turn. Ordered: any assistant tool-turns first (each bundling its tool calls with their executed responses), followed by the final plain-text assistant reply. nullable: true state: type: string description: Opaque encrypted state to carry forward to the next turn. nullable: true is_final: type: boolean description: >- True when the agent has signalled end-of-conversation (e.g. finishTheConversation). chat_id: type: string description: >- Server-issued correlation id. Stable across turns. Surfaced for logging. nullable: true call_id: type: string description: >- Internal Blackbox call id for this conversation. Stable across turns. Referenced in completion webhooks, the persisted call result, and audit events. nullable: true additionalProperties: false description: Result of a single stateless text-chat completion turn. TextChatErrorDto: type: object properties: code: type: string description: >- Stable machine-readable error code. See DashaAI.BlackBox.AgentAPI.TextChat.Dtos.TextChatErrorCodes. nullable: true message: type: string description: Human-readable error message. Safe to surface to API consumers. nullable: true details: type: object additionalProperties: {} description: >- Optional structured details bag for debugging. Never contains secrets. nullable: true additionalProperties: false description: Typed error envelope returned on failure of a text-chat completion call. ProblemDetails: type: object properties: type: type: string nullable: true title: type: string nullable: true status: type: integer format: int32 nullable: true detail: type: string nullable: true instance: type: string nullable: true additionalProperties: {} AgentRefDto: type: object properties: id: type: string description: >- Reference id of an existing agent owned by the calling organization. Set this OR DashaAI.BlackBox.AgentAPI.TextChat.Dtos.AgentRefDto.Config, not both. nullable: true config: allOf: - $ref: '#/components/schemas/AgentConfigDto' description: >- Complete configuration defining how the agent behaves during conversations. Includes voice synthesis settings, language model configuration, speech recognition, outbound calling capabilities, event notifications, advanced features, external service connections, and available tools. This is the core configuration that controls all aspects of agent behavior and capabilities. nullable: true additionalProperties: false description: >- Reference to the agent driving a text-chat completion — either an id of an existing agent or a full DashaAI.BlackBox.DataAccess.DTOs.Agent.AgentConfigDto snapshot supplied inline (useful for testing). Exactly one of DashaAI.BlackBox.AgentAPI.TextChat.Dtos.AgentRefDto.Id and DashaAI.BlackBox.AgentAPI.TextChat.Dtos.AgentRefDto.Config must be set. ChatMessageDto: type: object properties: role: type: string description: One of "user", "assistant", "system". nullable: true content: type: string description: >- Plain-text message content. Null when the assistant message only contains tool calls. nullable: true tool_calls: type: array items: $ref: '#/components/schemas/ChatToolCallDto' description: >- Tool calls produced by the assistant this turn, bundled with their responses. nullable: true time: type: string description: >- Real wall-clock time of the message. The server stamps the messages it produces and echoes back any client-supplied value; it is used verbatim for the call transcript / post-call analysis instead of a synthesized clock. One timestamp per chat message is enough — chat messages are point-in-time events, unlike voice utterances. Optional. format: date-time nullable: true additionalProperties: false description: >- Wire-format text-chat message. Supports user / assistant / system roles. Tool invocations are bundled — a single assistant message carries the tool call(s) alongside their executed response(s) via DashaAI.BlackBox.AgentAPI.TextChat.Dtos.ChatMessageDto.ToolCalls. AgentConfigDto: required: - llmConfig - primaryLanguage - ttsConfig type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true primaryLanguage: minLength: 1 type: string description: >- Primary language for the agent's conversations. Uses language codes like "en-US", "es-ES", "fr-FR". This determines the default language for speech synthesis and recognition. ttsConfig: allOf: - $ref: '#/components/schemas/TtsConfig' description: >- Text-to-speech configuration controlling how the agent's voice sounds. Includes voice provider selection, voice ID, speed, stability, and pronunciation dictionary settings. Defines the agent's vocal characteristics during conversations. llmConfig: allOf: - $ref: '#/components/schemas/LlmConfig' description: >- Language model configuration defining the agent's intelligence and conversation behavior. Includes model selection, system prompts, temperature, token limits, and conversation flow settings. Controls how the agent understands and responds to user input. sttConfig: allOf: - $ref: '#/components/schemas/SttConfig' description: >- Speech-to-text configuration for converting user speech to text. Includes provider selection and recognition settings. Controls how the agent understands spoken input from users. default: version: v1 vendor: Auto vendorSpecificOptions: {} nullable: true outboundSipConfig: oneOf: - $ref: '#/components/schemas/OutboundSipConfig' - $ref: '#/components/schemas/OutboundSipConfigV2' description: >- SIP configuration for making outbound calls. Includes phone number selection and trunk credentials. Required when the agent needs to initiate calls to phone numbers. default: null nullable: true resultWebhook: allOf: - $ref: '#/components/schemas/WebHook' description: >- Webhook URL to receive call results when conversations complete. The system will send a POST request with conversation outcome, transcript, and metadata to this URL. default: null nullable: true features: allOf: - $ref: '#/components/schemas/AgentFeatures' description: >- Advanced features for the agent including RAG (knowledge base retrieval), call transfers (warm or cold), ambient background sounds, interruption handling, and recording options. Enables specialized agent capabilities beyond basic conversation. default: version: v1 postCallAnalysis: [] talkFirst: version: v1 delay: 1 isEnabled: false interruptible: true maxDuration: version: v1 maxDurationSeconds: 3600 maxDurationInTransferSeconds: 7200 isEnabled: true ivrDetection: version: v1 isEnabled: true enabledForChat: false enabledForInbound: false enabledForOutbound: true enabledForWebCall: false ivrNavigation: false ambientNoise: version: v1 isEnabled: false ambientNoiseLevel: 0.1 languageSwitching: version: v1 isEnabled: true fillers: version: v1 isEnabled: true strategy: type: static texts: - um backchannel: version: v1 isEnabled: false turnTaking: version: v1 vad: asap_v1 silenceManagement: version: v1 isEnabled: true maxReminderAttempts: 2 reminderSilenceThresholdSeconds: 5 endWhenReminderLimitExceeded: true preConversationMedia: version: v1 isEnabled: false enabledForInbound: false enabledForOutbound: false enabledForWebCall: false noiseSuppression: version: v1 isEnabled: false type: none noiseSuppressionLevel: 75 recordWriteStrategy: raw nullable: true mcpConnections: type: array items: $ref: '#/components/schemas/McpConnection' description: >- Model Context Protocol connections for integrating external services. Allows the agent to connect to MCP servers via SSE or HTTP to access external tools and data sources. default: [] nullable: true startWebhook: allOf: - $ref: '#/components/schemas/StartWebhook' description: >- Webhook URL called when a call starts, before the conversation begins. Allows you to inject dynamic data or configuration at call initiation time. default: null nullable: true tools: type: array items: $ref: '#/components/schemas/LlmTool' description: >- Custom tools available to the agent during conversations. Each tool defines a function the agent can call with parameters and expected output. The agent will use these tools when needed to accomplish tasks beyond conversation. default: [] nullable: true builtInTools: type: array items: $ref: '#/components/schemas/BuiltInToolConfig' description: >- Built-in tools configuration with additional instructions for when and how to use them. Currently includes "finishTheConversation" tool with custom instructions that guide the agent on when to end calls gracefully. These are pre-defined system tools with configurable usage instructions. default: [] nullable: true additionalProperties: false description: >- Complete configuration defining how the agent behaves during conversations. Includes voice synthesis settings, language model configuration, speech recognition, outbound calling capabilities, event notifications, advanced features, external service connections, and available tools. This is the core configuration that controls all aspects of agent behavior and capabilities. ChatToolCallDto: type: object properties: id: type: string description: >- Tool call identifier propagated through the OpenAI round-trip. Required. nullable: true name: type: string description: >- Tool name as registered with the agent (built-in, agent-defined function, or MCP). nullable: true arguments: description: >- Tool arguments emitted by the LLM, as a JSON object. May be omitted for tools with no parameters. response: description: >- Tool response. Set after the server executes the tool. Null while the request is still in flight (only happens internally — wire-format messages from the client always include the response). nullable: true additionalProperties: false description: >- Combined tool call: the LLM-generated request and the executed response in a single object. Wire-format choice — internally the orchestrator translates to OpenAI's split assistant/tool format. TtsConfig: required: - vendor - voiceId type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true voiceId: minLength: 1 type: string description: >- Voice identifier from the selected vendor. Specifies which voice the agent uses for speech synthesis. Each vendor has different voice IDs representing distinct voice characteristics like gender, accent, and speaking style. vendor: minLength: 1 type: string description: >- Text-to-speech vendor providing the voice synthesis service. Identifies which TTS provider to use such as ElevenLabs, Cartesia, Dasha, Inworld, Lmnt, or Google. Different vendors offer different voice qualities, latency characteristics, and pricing. model: type: string description: >- TTS model identifier for vendors that support multiple model versions. Allows selecting specific synthesis models or quality tiers when available from the vendor. default: null nullable: true speed: type: number description: >- Speech speed multiplier controlling how fast the agent speaks. Values above 1.0 increase speaking rate, values below 1.0 decrease it. Affects the overall pace of conversation and can be adjusted to match user preferences or content requirements. format: double default: 1 nullable: true speedAdjustment: allOf: - $ref: '#/components/schemas/SpeedAdjustmentSettings' description: >- Speed adjustment settings controlling whether users can request the agent to speak faster or slower during conversations. Allows dynamic speed changes based on user feedback. default: version: v1 strategy: OnRequest nullable: true vendorSpecificOptions: type: object additionalProperties: {} description: >- Vendor-specific TTS options that are unique to particular speech synthesis providers. Allows passing custom parameters for advanced voice configuration that isn't covered by standard settings. nullable: true responsiveness: maximum: 1 minimum: 0 type: number description: >- Response timing control determining how quickly the agent begins speaking after the user finishes. Range is 0 to 1, where higher values make the agent respond faster and lower values add more deliberate pauses. Affects the conversational rhythm and turn-taking behavior. format: double default: null nullable: true pronunciationDictionary: allOf: - $ref: '#/components/schemas/PronunciationDictionaryReference' description: >- Reference to a pronunciation dictionary for customizing how specific words or phrases are pronounced. Allows correcting mispronunciations, defining custom pronunciations for brand names, or adjusting phonetic rendering for domain-specific terminology. default: null nullable: true additionalProperties: false description: >- Text-to-speech configuration defining how the agent's voice sounds and behaves. Specifies the voice provider, voice selection, speaking speed, response timing, and pronunciation customization. Controls the agent's vocal characteristics and speaking patterns during conversations. LlmConfig: required: - model - prompt - vendor type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true model: minLength: 1 type: string description: >- Language model identifier. Specifies which model to use from the configured vendor. vendor: minLength: 1 type: string description: Model vendor or provider identifier. prompt: minLength: 1 type: string description: >- System prompt that defines the agent's personality, role, and behavior instructions. This prompt guides how the agent responds and interacts during conversations. Can reference variables from AdditionalData using {{variableName}} syntax. apiKey: type: string description: >- API key for authenticating with the language model provider. Used to authorize requests to the vendor's API. default: null nullable: true endpoint: type: string description: >- Custom API endpoint URL for the language model. Use when connecting to self-hosted models or alternative API gateways instead of the default vendor endpoint. default: null nullable: true options: allOf: - $ref: '#/components/schemas/LlmOptions' description: >- Advanced language model parameters including temperature, max tokens, top P, frequency/presence penalties, stop sequences, trailing prompt, reasoning mode, and retry settings. Controls the model's creativity, response length, and behavior characteristics. default: null nullable: true vendorSpecificOptions: type: object additionalProperties: {} description: >- Vendor-specific options and parameters that are unique to particular LLM providers. Allows passing custom configuration that isn't covered by standard options. nullable: true toolScopes: type: array items: type: string description: >- Tool scope restrictions for limiting which tools the language model can access. When specified, the model can only use tools matching these scope identifiers. If not specified, all configured tools are available. default: null nullable: true additionalProperties: false description: >- Language model configuration defining the agent's intelligence and conversation behavior. Specifies which model to use, the system prompt, API credentials, and advanced options that control how the agent understands and responds to user input. SttConfig: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true vendor: allOf: - $ref: '#/components/schemas/SttVendor' description: >- Speech recognition vendor to use for transcribing user speech. Can be a specific provider like Deepgram or Microsoft, or Auto to let the system choose the best option automatically. default: Auto vendorSpecificOptions: type: object additionalProperties: {} description: >- Vendor-specific STT options that are unique to particular speech recognition providers. Allows passing custom parameters for advanced recognition configuration such as profanity filtering, punctuation, or language-specific settings. nullable: true keywords: type: array items: $ref: '#/components/schemas/SttKeyword' description: >- List of keywords or phrases that the speech recognition should prioritize or recognize more accurately. Helps improve transcription accuracy for domain-specific terminology, product names, or technical jargon that may not be in standard dictionaries. Each keyword can have an optional weight to control recognition priority. default: null nullable: true additionalProperties: false description: >- Speech-to-text configuration defining how the agent transcribes user speech into text. Specifies the speech recognition vendor, custom keywords for improved accuracy, and vendor-specific options. Controls the accuracy and behavior of speech recognition during conversations. OutboundSipConfig: required: - authUser - server type: object properties: type: enum: - v1 type: string description: >- Configuration version type. Always "v1" for inline credentials configuration. default: v1 nullable: true readOnly: true server: minLength: 1 type: string description: >- SIP server hostname or IP address with optional port as defined in RFC 3261. Format is hostname:port (e.g., "sip.example.com:5060"). authUser: minLength: 1 type: string description: >- Authentication user identifier as defined in RFC 3261. Different providers may call this "username", "account", "user ID", or other names depending on the SIP provider. domain: type: string description: >- SIP domain for authentication realm as defined in RFC 3261. Used in SIP authentication challenges. default: null nullable: true password: type: string description: >- Password for SIP authentication as defined in RFC 3261. Used for authenticating outbound calls. default: null nullable: true fromUser: type: string description: >- From header user portion for outbound calls. Determines what user identifier appears in the SIP From header when making calls. default: null nullable: true transport: allOf: - $ref: '#/components/schemas/TransportProtocol' description: >- Transport protocol for SIP communication. Supported values are UDP (default), TCP, or TLS for encrypted connections. default: udp nullable: true cpsLimit: type: integer description: >- Calls per second limit for rate limiting outbound calls. Prevents exceeding provider rate limits. format: int32 default: null nullable: true cpsKey: type: string description: >- Calls per second rate limiting key. Used to group multiple configurations under the same rate limit when specified. default: null nullable: true displayName: type: string description: >- Display name for caller ID. Appears in the From header display name when making outbound calls. default: null nullable: true additionalProperties: false description: >- Outbound SIP configuration V1 with inline credentials. Contains complete SIP trunk settings including server address, authentication credentials, and transport settings as defined in RFC 3261. OutboundSipConfigV2: required: - phoneNumbers - type type: object properties: type: minLength: 1 enum: - v2 type: string description: >- Configuration version type. Always "v2" for phone number reference configuration. readOnly: true phoneNumbers: type: array items: $ref: '#/components/schemas/OutboundSipPhoneNumber' description: >- List of phone number references for outbound calling. Currently supports only one phone number in the list. Each reference points to a configured SIP phone number that contains the trunk credentials and settings. additionalProperties: false description: >- Outbound SIP configuration V2 that references phone numbers instead of containing inline credentials. References phone numbers configured via the SIP phone numbers API, which contain the actual trunk credentials and settings. This configuration is resolved to V1 format at runtime by looking up the phone number details. WebHook: required: - url type: object properties: url: type: string description: >- Webhook endpoint URL to send HTTP requests to. The system will POST data to this URL when the webhook is triggered. format: uri headers: type: object additionalProperties: type: string description: >- Custom HTTP headers to include with webhook requests. Useful for authentication, API keys, or passing additional metadata to the receiving endpoint. nullable: true customSettings: allOf: - $ref: '#/components/schemas/CustomWebhookSettings' description: >- Custom webhook request settings including HTTP method, query parameters, and body format configuration. Allows full control over how the webhook request is constructed and sent. default: null nullable: true additionalProperties: false description: >- Webhook configuration for HTTP callbacks. Defines the URL endpoint to call, custom HTTP headers, and request customization settings for sending event data or tool invocations to external systems. AgentFeatures: type: object properties: version: enum: - v1 type: string description: Feature configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true transfer: oneOf: - $ref: '#/components/schemas/ColdTransfer' - $ref: '#/components/schemas/HttpTransfer' - $ref: '#/components/schemas/WarmTransfer' description: >- Call transfer configuration. Supports warm transfer (agent stays on call), cold transfer (agent hangs up), or HTTP transfer (external API handles transfer). Enables the agent to transfer calls to other numbers or systems. default: null nullable: true postCallAnalysis: type: array items: $ref: '#/components/schemas/PostCallAnalysis' description: >- Post-call analysis configurations for extracting insights after conversations complete. Multiple analysis tasks can be configured to process transcripts and generate structured data. default: [] nullable: true talkFirst: allOf: - $ref: '#/components/schemas/TalkFirst' description: >- Controls whether the agent speaks first when the call connects. When enabled, the agent greets the user immediately rather than waiting for the user to speak first. default: version: v1 delay: 1 isEnabled: false interruptible: true nullable: true maxDuration: allOf: - $ref: '#/components/schemas/MaxDuration' description: >- Maximum duration limit for conversations. The call will be automatically ended if it exceeds the configured time limit. default: version: v1 maxDurationSeconds: 3600 maxDurationInTransferSeconds: 7200 isEnabled: true nullable: true ivrDetection: allOf: - $ref: '#/components/schemas/IvrDetection' description: >- IVR (Interactive Voice Response) detection for identifying automated phone systems. Helps the agent recognize and respond appropriately to menu prompts and automated messages. default: version: v1 isEnabled: true enabledForChat: false enabledForInbound: false enabledForOutbound: true enabledForWebCall: false ivrNavigation: false nullable: true ambientNoise: allOf: - $ref: '#/components/schemas/AmbientNoise' description: >- Ambient background noise configuration. Adds realistic background sounds like cafe ambiance or office noise to make the agent sound more natural and human-like. default: version: v1 isEnabled: false ambientNoiseLevel: 0.1 nullable: true rag: allOf: - $ref: '#/components/schemas/RAGFeature' description: >- RAG (Retrieval-Augmented Generation) feature for knowledge base integration. Enables the agent to search and retrieve information from connected knowledge bases during conversations. default: null nullable: true languageSwitching: allOf: - $ref: '#/components/schemas/LanguageSwitchingFeature' description: >- Language switching capability allowing the agent to switch between multiple languages during the call. Enables the agent to communicate in different languages within the same conversation. default: version: v1 isEnabled: true nullable: true fillers: allOf: - $ref: '#/components/schemas/FillerConfig' description: >- Filler words and phrases configuration. Controls natural speech patterns like "um", "uh", "let me see" to make the agent sound more conversational and human-like. default: version: v1 isEnabled: true strategy: type: static texts: - um nullable: true backchannel: allOf: - $ref: '#/components/schemas/Backchannel' description: >- Backchannel responses configuration. Controls brief acknowledgments like "mm-hmm", "I see", "right" that the agent uses while listening to show active engagement. default: version: v1 isEnabled: false nullable: true turnTaking: allOf: - $ref: '#/components/schemas/TurnTaking' description: >- Turn taking configuration for the main customer call. Controls which DashaScript VAD mode is passed to the initial connect operation. default: version: v1 vad: asap_v1 nullable: true silenceManagement: allOf: - $ref: '#/components/schemas/SilenceManagement' description: >- Silence management configuration. Controls how the agent handles pauses and silence during conversations including timeout behavior and prompting strategies. default: version: v1 isEnabled: true maxReminderAttempts: 2 reminderSilenceThresholdSeconds: 5 endWhenReminderLimitExceeded: true nullable: true preConversationMedia: allOf: - $ref: '#/components/schemas/PreConversationMedia' description: >- Pre-conversation media playback. Plays an audio file immediately when the call connects, before the conversation begins. The media plays once and is non-interruptible. Can be enabled independently for inbound, outbound, and web calls. default: version: v1 isEnabled: false enabledForInbound: false enabledForOutbound: false enabledForWebCall: false nullable: true noiseSuppression: allOf: - $ref: '#/components/schemas/NoiseSuppression' description: >- Noise suppression configuration for filtering background noise from call audio. Supports "none" (disabled) or "krisp" (Krisp engine, additional fee applies). Also controls the noise suppression level and recording write strategy. default: version: v1 isEnabled: false type: none noiseSuppressionLevel: 75 recordWriteStrategy: raw nullable: true additionalProperties: false description: >- Advanced features configuration for the agent. Controls specialized capabilities including call transfers, knowledge base retrieval, ambient sounds, conversation management, language switching, and post-call analysis. McpConnection: required: - name - serverUrl type: object properties: name: minLength: 1 type: string description: >- Display name for this MCP connection. Used to identify the connection in logs and configuration. serverUrl: type: string description: >- MCP server URL to connect to. The URL format depends on the transport protocol (SSE uses HTTP/HTTPS endpoints, StreamableHTTP uses HTTP endpoints). format: uri authentication: oneOf: - $ref: '#/components/schemas/ApiKeyAuthentication' - $ref: '#/components/schemas/BearerAuthentication' - $ref: '#/components/schemas/NoAuthentication' description: >- Authentication configuration for the MCP server. Supported types include none, API key, bearer token, and custom authentication schemes. default: null nullable: true customHeaders: type: object additionalProperties: type: string description: >- Custom HTTP headers to include with each request to the MCP server. Useful for passing additional metadata or configuration to the server. nullable: true isEnabled: type: boolean description: >- Whether this MCP connection is enabled. When disabled, the agent will not connect to this MCP server or use its tools. default: true nullable: true description: type: string description: >- Optional description explaining the purpose of this MCP connection or what services it provides. default: null nullable: true transport: allOf: - $ref: '#/components/schemas/McpConnectionTransport' description: >- Transport protocol for MCP communication. Supported values are SSE (Server-Sent Events for persistent connections) and StreamableHTTP (for request-response patterns). default: SSE nullable: true blackListTools: type: array items: type: string description: >- List of tool names to exclude from this MCP connection. Tools in this list will not be available to the agent even if the MCP server provides them. Useful for restricting access to specific tools. default: null nullable: true whiteListTools: type: array items: type: string description: >- List of tool names to allow from this MCP connection. When specified, only tools in this list will be available to the agent. If not specified, all tools from the server are available (subject to BlackListTools). default: null nullable: true additionalProperties: false description: >- Model Context Protocol connection configuration for integrating external services with the agent. Defines how to connect to an MCP server including server URL, authentication, transport protocol, custom headers, and tool filtering. StartWebhook: required: - webhook type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true webhook: allOf: - $ref: '#/components/schemas/WebHook' description: >- Webhook configuration defining the endpoint, headers, and request settings for start-of-conversation notifications. The system sends conversation initiation data to this webhook and waits for a response. fallbackResponse: oneOf: - $ref: '#/components/schemas/StartWebHookResponseAccept' - $ref: '#/components/schemas/StartWebHookResponseReject' description: >- Fallback response to use when the webhook call fails or times out. Determines whether to accept or reject the conversation if the webhook endpoint is unavailable or returns an error. Can be configured to accept the call with default settings or reject it. default: accept: false reasonMessage: Webhook for started webhook failed nullable: true isEnabledForInbound: type: boolean description: >- Enables the start webhook for inbound phone calls. When enabled, the webhook is triggered when someone calls one of your configured phone numbers, allowing you to accept or reject incoming calls based on external logic. default: false nullable: true isEnabledForOutbound: type: boolean description: >- Enables the start webhook for outbound phone calls. When enabled, the webhook is triggered before making an outbound call, allowing you to modify call parameters or cancel the call based on external conditions. default: false nullable: true isEnabledForWidget: type: boolean description: >- Enables the start webhook for widget-based conversations. When enabled, the webhook is triggered when a user initiates a conversation through your web widget, allowing you to accept or reject widget interactions. default: false nullable: true timeout: type: number description: >- Maximum time in seconds to wait for the webhook response. If the webhook doesn't respond within this time, the fallback response is used. Controls how long the system waits before proceeding with or rejecting the conversation. format: double default: 10 nullable: true additionalProperties: false description: >- Webhook triggered at the start of a conversation or call. Sends a notification to external systems when a new interaction begins, allowing them to accept or reject the call, modify conversation parameters, or perform initialization logic. Supports inbound calls, outbound calls, and widget-based conversations. LlmTool: required: - description - name - schema - webhook type: object properties: name: minLength: 1 type: string description: >- Tool name identifier. Used by the language model to reference this tool when deciding which function to call. schema: type: object additionalProperties: {} description: >- JSON Schema defining the tool's input parameters. Describes what arguments the tool accepts, their types, and which parameters are required. The language model uses this schema to generate valid function calls. webhook: allOf: - $ref: '#/components/schemas/WebHook' description: >- Webhook endpoint to call when the tool is invoked. The system sends a POST request with the tool parameters to this URL and returns the response to the language model. description: minLength: 1 type: string description: >- Tool description explaining what the tool does and when to use it. The language model uses this description to decide whether this tool is appropriate for the current conversation context. fallBackResult: description: >- Fallback result to return if the webhook call fails. When specified, the agent continues the conversation using this result instead of reporting an error to the user. default: null nullable: true timeoutSeconds: type: integer description: >- Maximum time in seconds to wait for the webhook response. If the webhook doesn't respond within this time, the call is considered failed and the fallback result is used if available. format: int32 default: null nullable: true retryCount: type: integer description: >- Number of retry attempts if the webhook call fails. Helps handle transient network errors or temporary service unavailability. format: int32 default: null nullable: true additionalProperties: false description: >- Custom tool definition for language model function calling. Describes a function that the agent can invoke during conversations to perform actions or retrieve information through webhook calls. BuiltInToolConfig: required: - name type: object properties: name: minLength: 1 type: string description: >- Built-in tool name. Currently supported tool is "finishTheConversation" for ending calls gracefully. additionalInstructions: type: string description: >- Additional instructions for when and how to use this tool. Guides the agent on the appropriate circumstances to invoke the tool during conversations. default: null nullable: true additionalProperties: false description: >- Configuration for a built-in tool with custom usage instructions. Built-in tools are pre-defined system tools that don't require custom implementation, and this configuration allows you to specify when and how the agent should use them. SpeedAdjustmentSettings: required: - strategy type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true strategy: allOf: - $ref: '#/components/schemas/SpeedAdjustment' description: >- Speed adjustment strategy determining when and how the agent can change its speaking rate. Controls whether speed adjustments are disabled entirely or enabled on user request. additionalProperties: false description: >- Speech speed adjustment configuration controlling whether the agent can change its speaking rate during conversations. Determines if users can request the agent to speak faster or slower to match their comprehension preferences. PronunciationDictionaryReference: required: - hash - id type: object properties: id: minLength: 1 type: string description: >- Unique identifier of the pronunciation dictionary to use. References a dictionary that has been created and uploaded to the system containing custom phonetic rules and pronunciation mappings. hash: minLength: 1 type: string description: >- Content hash of the pronunciation dictionary for version tracking and cache invalidation. Ensures the correct version of the dictionary is applied and allows detecting when dictionary content has been updated. additionalProperties: false description: >- Reference to a pronunciation dictionary for customizing text-to-speech phonetic rendering. Links to a dictionary that defines custom pronunciations for specific words, brand names, technical terms, or proper nouns. Helps correct TTS mispronunciations and ensure consistent pronunciation of domain-specific vocabulary. The hash enables version tracking to detect when dictionary content changes. LlmOptions: type: object properties: temperature: type: number description: >- Temperature controls randomness in responses. Higher values (e.g., 0.8-1.0) make output more creative and varied, while lower values (e.g., 0.1-0.3) make it more focused and deterministic. Range is typically 0.0 to 2.0. format: double default: null nullable: true maxTokens: type: integer description: >- Maximum number of tokens the model can generate in a response. Limits the length of the agent's replies to control costs and response times. format: int32 default: null nullable: true topP: type: number description: >- Top P (nucleus sampling) controls diversity by considering only the most probable tokens whose cumulative probability reaches this threshold. Range is 0.0 to 1.0, where lower values make responses more focused. format: double default: null nullable: true frequencyPenalty: type: number description: >- Frequency penalty reduces repetition of tokens based on how often they've appeared in the response. Higher values (e.g., 0.5-1.0) discourage repetition more strongly. Range is typically -2.0 to 2.0. format: double default: null nullable: true presencePenalty: type: number description: >- Presence penalty reduces repetition by penalizing tokens that have already appeared, regardless of frequency. Encourages the model to discuss new topics. Range is typically -2.0 to 2.0. format: double default: null nullable: true stop: type: array items: type: string description: >- Stop sequences that cause the model to stop generating tokens. When the model encounters any of these strings, it will immediately stop generating further output. default: null nullable: true trailingPrompt: type: string description: >- Trailing prompt appended to the conversation context before the model generates a response. Useful for providing last-minute instructions or context. default: null nullable: true reasoningMode: type: string description: >- Reasoning mode for models that support structured thinking. Controls how the model approaches problem-solving and response generation. default: null nullable: true maxRetries: type: integer description: >- Maximum number of retry attempts when API calls fail. Helps handle transient errors and rate limiting. format: int32 default: null nullable: true reasoningEffort: type: string description: >- Reasoning effort level for models that support configurable thinking depth. Controls how much computational effort the model applies to reasoning through complex problems. default: null nullable: true additionalProperties: false description: >- Advanced language model parameters controlling response generation behavior. These options fine-tune how the model produces text including creativity, randomness, repetition avoidance, and token limits. SttVendor: enum: - Deepgram - Microsoft - Auto type: string description: >- Speech-to-text vendor providing the speech recognition service. Identifies which STT provider to use for transcribing user speech into text. Different vendors offer varying accuracy, language support, and latency characteristics. SttKeyword: required: - keyword type: object properties: keyword: minLength: 1 type: string description: >- Word or phrase to prioritize during speech recognition. The STT engine gives preference to recognizing this keyword when transcribing user speech, improving accuracy for domain-specific vocabulary. weight: type: number description: >- Optional weight controlling how strongly the keyword should be prioritized. Higher weights increase the likelihood of the keyword being recognized over similar-sounding alternatives. When not specified, a default prioritization is applied. format: double nullable: true additionalProperties: false description: >- Speech recognition keyword hint for improving transcription accuracy of specific words or phrases. Tells the STT engine to prioritize recognition of domain-specific terminology, product names, technical jargon, or proper nouns that may not be in standard dictionaries. Helps reduce misrecognition of important context-specific vocabulary. TransportProtocol: enum: - udp - tcp - tls type: string description: >- Transport protocol for SIP communication. Defines how SIP messages are transmitted over the network. OutboundSipPhoneNumber: required: - phoneNumberId type: object properties: phoneNumberId: type: string description: >- Unique identifier of the SIP phone number configuration to use for outbound calls. References a phone number configured via the SIP phone numbers API. format: uuid additionalProperties: false description: >- Reference to a configured SIP phone number. Contains the phone number ID that points to the full phone number configuration including trunk credentials. CustomWebhookSettings: required: - httpMethod type: object properties: httpMethod: allOf: - $ref: '#/components/schemas/HttpMethod' description: >- HTTP method to use for the webhook request. Determines whether the request uses GET, POST, PUT, PATCH, or DELETE. POST is the most common choice for webhook notifications. queryParams: type: object additionalProperties: type: array items: type: string description: >- Query string parameters to append to the webhook URL. Each parameter can have multiple values. These parameters are URL-encoded and added to the request URL as key=value pairs separated by ampersands. nullable: true bodyFormat: oneOf: - $ref: '#/components/schemas/BodyTextFormat' - $ref: '#/components/schemas/BodyJSONFormat' - $ref: '#/components/schemas/BodyFormFormat' description: >- Format configuration for the request body. Specifies how the webhook payload is encoded and what Content-Type header is sent. Supports JSON, plain text, and form-encoded formats for compatibility with different endpoint requirements. default: null nullable: true additionalProperties: false description: >- Custom request settings for webhook HTTP calls. Allows full control over how webhook requests are constructed including the HTTP method, URL query parameters, and request body formatting. These settings enable integration with REST APIs and services that require specific request formats. ColdTransfer: allOf: - $ref: '#/components/schemas/TransferFeatureBase' - required: - endpointDestination - type type: object properties: type: enum: - cold type: string description: >- Call transfer type determining how the call is handed off to a human operator or different destination. Different transfer types control the handoff mechanism, agent involvement during transfer, and whether decisions are made dynamically via HTTP requests. readOnly: true endpointDestination: minLength: 1 type: string description: >- Transfer destination phone number or SIP address. Supports E.164 phone numbers like "+1234567890" or SIP URIs like "sip:1234567890@example.com". The call is redirected to this endpoint using SIP REFER when the transfer is initiated. transferRoutes: type: array items: $ref: '#/components/schemas/TransferRoute' description: >- Optional named transfer routes allowing the language model to choose between multiple destinations. Each route has a name, description, and list of endpoints. The model can select the appropriate route based on conversation context, enabling intelligent routing to different departments or specialists. default: null nullable: true holdMedia: allOf: - $ref: '#/components/schemas/HoldMediaConfig' description: >- Optional hold media configuration for playing audio during the transfer. When enabled, plays the specified media file on loop while the call is being transferred instead of using the agent's voice. Provides professional hold music or messages while the transfer completes. default: null nullable: true additionalProperties: false description: >- Cold transfer configuration using SIP REFER for immediate call handoff. The AI agent disconnects as soon as the transfer completes, connecting the user directly to the destination operator without remaining on the call. Uses the SIP REFER method according to RFC 3515 to instruct the SIP infrastructure to redirect the call. Suitable for simple handoffs where the agent doesn't need to provide context to the operator or remain involved in the conversation. HttpTransfer: allOf: - $ref: '#/components/schemas/TransferFeatureBase' - required: - type - webhook type: object properties: type: enum: - http type: string description: >- Call transfer type determining how the call is handed off to a human operator or different destination. Different transfer types control the handoff mechanism, agent involvement during transfer, and whether decisions are made dynamically via HTTP requests. readOnly: true webhook: allOf: - $ref: '#/components/schemas/WebHook' description: >- Webhook endpoint to call for transfer routing decisions. The system sends a POST request with conversation context and waits for a response specifying the transfer destination and type. The webhook should return a transfer decision in the response body indicating whether to perform a cold or warm transfer and to which endpoint. fallback: oneOf: - $ref: '#/components/schemas/ColdTransfer' - $ref: '#/components/schemas/WarmTransfer' description: >- Fallback transfer configuration to use if the webhook request fails, times out, or returns an error. Ensures calls can still be transferred even when the routing system is unavailable. Can specify either a cold or warm transfer as the fallback behavior. default: null nullable: true additionalProperties: false description: >- HTTP-based dynamic transfer configuration using webhook requests to determine routing. Sends a webhook POST request with the conversation context, allowing external systems to make intelligent routing decisions based on caller information, conversation content, or business logic. The webhook response specifies which transfer destination to use and whether to perform a cold or warm transfer. Enables advanced routing scenarios like skill-based routing, availability checking, or CRM integration for transfer decisions. WarmTransfer: allOf: - $ref: '#/components/schemas/TransferFeatureBase' - required: - endpointDestination - type type: object properties: type: enum: - warm type: string description: >- Call transfer type determining how the call is handed off to a human operator or different destination. Different transfer types control the handoff mechanism, agent involvement during transfer, and whether decisions are made dynamically via HTTP requests. readOnly: true sip: allOf: - $ref: '#/components/schemas/OptionalOutboundSipConfig' description: >- Optional SIP configuration override for placing the call to the operator. When specified, uses these SIP credentials instead of the agent's default outbound SIP settings for this specific transfer call. default: null nullable: true endpointDestination: minLength: 1 type: string description: >- Operator destination phone number or SIP address. Supports E.164 phone numbers like "+1234567890" or SIP URIs like "sip:operator@example.com". The agent calls this endpoint first before bridging with the user. interactionWithOperator: oneOf: - $ref: '#/components/schemas/SmartV1Behavior' - $ref: '#/components/schemas/StaticBehavior' description: >- Behavior controlling what the agent says to the operator when they answer. Can use static predefined phrases or smart AI-generated context based on the conversation. Allows providing the operator with relevant information about the caller's needs before connecting them. default: null nullable: true sayPhraseToCustomer: type: string description: >- Phrase to speak to the customer before bridging the call with the operator. Informs the user that they're being connected to a human agent. Can include reassurance or set expectations about the handoff. default: null nullable: true continueAfterOperatorDisconnected: type: boolean description: >- Controls whether the AI agent resumes the conversation if the operator disconnects or hangs up. When true, the agent continues talking to the user after the operator leaves. When false, the call ends when the operator disconnects. default: false nullable: true continueRecordingInTransfer: type: boolean default: true nullable: true transferRoutes: type: array items: $ref: '#/components/schemas/TransferRoute' default: null nullable: true operatorName: type: string default: null nullable: true continueRecognitionInTransfer: type: boolean default: false nullable: true callerIdMode: allOf: - $ref: '#/components/schemas/CallerIdMode' description: >- Determines which phone number is displayed to the transfer destination (human agent) during warm transfer. AgentPhoneNumber (default): Uses the AI agent's configured outbound SIP fromUser as the caller ID. UserPhoneNumber: Uses the user's phone number (job.endpoint) as the caller ID. When null, defaults to AgentPhoneNumber behavior. default: agentPhoneNumber nullable: true threewayMessaging: oneOf: - $ref: '#/components/schemas/ThreewaySmartV1Behavior' - $ref: '#/components/schemas/ThreewayStaticBehavior' description: >- Optional threeway messaging configuration. When configured, delivers a unified phrase to BOTH user and operator simultaneously before the bridge channel is executed. default: null nullable: true holdMedia: allOf: - $ref: '#/components/schemas/HoldMediaConfig' description: >- Optional hold media configuration for warm transfers. When enabled, plays media via media:// protocol instead of using the respond node to communicate with users during transfer. default: null nullable: true additionalProperties: false description: >- Warm transfer configuration for agent-assisted call handoff. The AI agent first calls the destination operator, can speak with them to provide context or a conversation summary, and then connects the user. Supports configurable behaviors for what the agent says to the operator and customer, three-way calling where the agent remains connected, and continuation of the conversation if the operator disconnects. Provides a more personalized handoff experience than cold transfers. PostCallAnalysis: required: - labels - name type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true name: maxLength: 100 minLength: 1 pattern: ^[a-zA-Z\-_][a-zA-Z0-9\-_]*$ type: string description: >- Name identifier for this analysis configuration. Must start with a letter, hyphen, or underscore and contain only letters, numbers, hyphens, and underscores. Used to identify this analysis in the results. labels: type: array items: $ref: '#/components/schemas/PostCallAnalysisLabel' description: >- List of data fields to extract from the conversation transcript. Each label defines a piece of information to identify and extract, such as customer sentiment, appointment times, order numbers, or custom business outcomes. The language model analyzes the conversation and populates these labels based on the transcript content. isEnabled: type: boolean description: >- Enables or disables this post-call analysis. When enabled, the system performs the configured analysis after each call completes and includes results in the completion webhook payload. default: true nullable: true additionalProperties: false description: >- Post-call analysis configuration for automatically extracting structured data from conversation transcripts after calls complete. Defines labels and fields to extract using language model analysis of the conversation. Enables automated classification, sentiment analysis, information extraction, and business outcome tracking from completed calls. Results are included in the completed webhook payload for integration with CRM systems, analytics platforms, or business intelligence tools. TalkFirst: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true phrase: type: string description: >- Greeting phrase the agent should speak immediately when the call connects. This is spoken before the normal conversation flow begins, allowing the agent to introduce itself or provide an opening statement on outbound calls. default: null nullable: true delay: type: number description: >- Delay in seconds before speaking the greeting phrase. Allows time for the call connection to stabilize or for the user to say hello first if desired. The agent waits this duration after call connection before speaking the configured phrase. format: double default: 1 nullable: true isEnabled: type: boolean description: >- Enables or disables the talk first feature. When enabled, the agent speaks the configured greeting phrase after the specified delay when the call connects. default: false nullable: true interruptible: type: boolean description: >- Controls whether the user can interrupt the greeting phrase. When true, the agent stops speaking the greeting if the user starts talking. When false, the agent completes the entire greeting phrase before listening for user input. default: true nullable: true additionalProperties: false description: >- Talk first feature for having the agent speak an initial greeting immediately when the call connects, before waiting for user input. Useful for outbound calls or situations where the agent should introduce itself proactively. The greeting can be delayed and configured to be interruptible by user speech. MaxDuration: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true maxDurationSeconds: type: integer description: >- Maximum duration in seconds for normal conversations before the call is automatically ended. Prevents calls from running indefinitely and helps control costs and resource usage. format: int32 default: 3600 nullable: true maxDurationInTransferSeconds: type: integer description: >- Maximum duration in seconds for conversations that have been transferred. Allows longer durations for transferred calls since they may involve more complex interactions or human operators. The call is automatically ended when this limit is exceeded during a transfer. format: int32 default: 7200 nullable: true isEnabled: type: boolean description: >- Enables or disables maximum duration enforcement. When enabled, calls are automatically terminated when they exceed the configured duration limits. default: true nullable: true additionalProperties: false description: >- Maximum duration limits for conversations to prevent excessively long calls. Automatically ends conversations that exceed the configured time limits, with separate thresholds for normal conversations and transferred calls. Useful for cost control, resource management, and preventing stuck calls. IvrDetection: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Master enable flag for IVR and answering machine detection. When enabled, the system attempts to detect whether calls are answered by humans or machines and behaves according to the configured settings. default: true nullable: true enabledForChat: type: boolean description: >- Enables answering machine detection for chat conversations. When enabled, the system attempts to detect automated chat responses or bot interactions. default: false nullable: true enabledForInbound: type: boolean description: >- Enables answering machine detection for inbound phone calls. When enabled, the system detects if the caller is a human or an automated system calling your phone numbers. default: false nullable: true enabledForOutbound: type: boolean description: >- Enables answering machine detection for outbound phone calls. When enabled, the system detects whether a human or voicemail/IVR system answered the call, allowing the agent to leave messages or navigate menus as configured. default: true nullable: true enabledForWebCall: type: boolean description: >- Enables answering machine detection for web-based voice calls. When enabled, the system detects automated responses in web call scenarios. default: false nullable: true ivrNavigation: type: boolean description: >- Enables IVR menu navigation using DTMF tone sending. When enabled, the agent attempts to navigate through automated phone menus by sending touch-tone digits before giving up on the call. Useful for reaching human operators through phone trees on outbound calls. default: false nullable: true experimental: type: object additionalProperties: {} description: Experimental features default: null nullable: true behavior: oneOf: - $ref: '#/components/schemas/SmartV1IvrDetectionBehavior' - $ref: '#/components/schemas/StaticIvrDetectionBehavior' description: >- Behavior configuration controlling what the agent says or does when an answering machine or IVR is detected. Static behaviors use predefined phrases, while smart behaviors use the language model to generate contextual responses based on the conversation. default: null nullable: true staticPhraseForIVR: type: string description: >- Deprecated static phrase for answering machine messages. Use the Behavior property instead for more flexible message configuration. This field supports variable interpolation using {{variableName}} syntax. default: null nullable: true deprecated: true additionalProperties: false description: >- IVR and answering machine detection configuration controlling how the agent behaves when calls are answered by automated systems or voicemail. Detects whether a human or machine answered the call and can leave messages on answering machines, navigate through IVR menus using DTMF tones, or end calls that reach voicemail. Helps optimize outbound calling campaigns by handling automated responses appropriately. AmbientNoise: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables ambient noise playback. When enabled, the configured background sound or noise level is mixed with the agent's voice output. default: false nullable: true ambientNoiseLevel: type: number description: >- Volume level for the ambient noise relative to the agent's voice. Controls how loud the background noise is compared to the agent's speech. Lower values create subtle background ambience, higher values make the background more prominent. format: double default: 0.1 nullable: true backgroundSound: type: string description: >- Identifier for the background sound to play. Specifies which ambient sound effect or noise pattern to use, such as office sounds, cafe ambience, or generic background noise. default: null nullable: true additionalProperties: false description: >- Ambient noise configuration for adding background sounds to the agent's audio output. Enhances naturalness by simulating realistic call environments with subtle background noise or specific sound effects. Can make the agent sound more human-like by avoiding completely silent backgrounds. RAGFeature: required: - kbLinks type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables RAG functionality. When enabled, the agent can search linked knowledge bases during conversations to retrieve relevant information for answering user questions with accurate, contextual responses. default: false nullable: true kbLinks: type: array items: $ref: '#/components/schemas/KBLink' description: >- List of knowledge base references that the agent can search. Each link identifies a knowledge base containing documents, FAQs, or reference materials. The agent can query these knowledge bases during conversations to augment its responses with retrieved information. additionalProperties: false description: >- Retrieval-Augmented Generation (RAG) feature configuration for connecting the agent to knowledge bases. Enables the agent to search and retrieve information from document repositories during conversations to answer questions with up-to-date, domain-specific knowledge. Links to one or more knowledge bases that have been created and populated with documents, articles, or reference materials. LanguageSwitchingFeature: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables the language switching capability. When enabled, the agent can switch languages mid-conversation in response to user requests to speak in a different language. default: true nullable: true additionalProperties: false description: >- Language switching feature allowing the agent to switch between multiple languages during the call based on user requests. When enabled, the agent can change the conversation language when the user explicitly asks to speak in a different language, enabling multilingual conversations within a single call. FillerConfig: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables filler word insertion. When enabled, the agent occasionally uses the configured filler phrases during conversations to sound more natural and less robotic. default: true nullable: true strategy: oneOf: - $ref: '#/components/schemas/FillerStaticStrategy' description: >- Strategy for selecting and inserting filler words. Static strategies use a predefined list of filler phrases. The strategy controls which fillers to use and how often they appear in the agent's speech. default: type: static texts: - um nullable: true additionalProperties: false description: >- Filler words configuration for making the agent's speech sound more natural and human-like. Fillers are verbal pauses like "um", "uh", "let me see", or "well" that the agent inserts while thinking or processing. These small hesitations make conversations feel more natural by mimicking human speech patterns instead of perfectly fluent robotic responses. Backchannel: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables backchannel acknowledgments. When enabled, the agent produces listening cues while the user speaks to indicate attention and understanding. default: false nullable: true behavior: oneOf: - $ref: '#/components/schemas/SmartV1BackchannelBehavior' - $ref: '#/components/schemas/StaticAdvancedBackchannelBehavior' - $ref: '#/components/schemas/StaticBackchannelBehavior' description: >- Behavior configuration controlling when and how the agent produces backchannels. Static behaviors use configurable timing properties to determine when to insert acknowledgments. Smart behaviors use intelligent detection to produce backchannels at natural conversation points. default: null nullable: true additionalProperties: false description: >- Backchannel feature configuration for having the agent produce acknowledgment sounds while the user is speaking. Backchannels are verbal cues like "uh-huh", "mm-hmm", or "I see" that signal active listening and engagement. Makes conversations feel more natural and interactive by mimicking human listening behavior. TurnTaking: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true vad: allOf: - $ref: '#/components/schemas/TurnTakingVad' description: >- Voice activity detection mode passed to the main call connect options. default: asap_v1 additionalProperties: false description: Turn taking configuration for the main customer call. SilenceManagement: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables silence management. When enabled, the agent sends reminders when the user is silent and can automatically end conversations after inactivity thresholds are exceeded. When disabled, the agent waits indefinitely for user input without prompting or terminating. default: true nullable: true maxReminderAttempts: type: integer description: >- Maximum number of reminder attempts to make when the user is silent. After this limit is reached, behavior is controlled by EndWhenReminderLimitExceeded to determine whether to end the call or stop reminding. format: int32 default: 2 nullable: true reminderSilenceThresholdSeconds: type: number description: >- Silence duration in seconds after which a reminder is triggered. The agent waits this long for user input before prompting the user to ensure they are still engaged in the conversation. format: double default: 5 nullable: true endWhenReminderLimitExceeded: type: boolean description: >- Controls whether to end the conversation when maximum reminder attempts have been exceeded. When true, the call terminates after all reminders are exhausted. When false, the agent stops sending reminders but keeps the conversation open, waiting indefinitely for user input. default: true nullable: true endAfterSilenceThresholdSeconds: type: number description: >- Absolute silence duration in seconds after which the conversation is automatically ended, regardless of reminder attempts. Provides a hard timeout for user inactivity. When not specified, only reminder-based termination applies. format: double default: null nullable: true additionalProperties: false description: >- Silence management configuration controlling agent behavior when the user becomes silent during a conversation. Defines reminder attempts and conversation termination thresholds based on user inactivity. Prevents conversations from hanging indefinitely by prompting silent users or ending the call after extended silence. PreConversationMedia: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: Enables or disables pre-conversation media playback. default: false nullable: true mediaFilename: type: string description: >- Filename of the uploaded media file to play before the conversation starts. Must reference a media file that has been uploaded to the system. default: null nullable: true enabledForInbound: type: boolean description: Enables pre-conversation media for inbound phone calls. default: false nullable: true enabledForOutbound: type: boolean description: Enables pre-conversation media for outbound phone calls. default: false nullable: true enabledForWebCall: type: boolean description: Enables pre-conversation media for web-based voice calls. default: false nullable: true additionalProperties: false description: >- Pre-conversation media playback configuration. When enabled, plays an audio file immediately when the call connects, before the conversation begins. The media plays once and is non-interruptible. Can be enabled independently for inbound, outbound, and web calls. NoiseSuppression: type: object properties: version: enum: - v1 type: string description: Configuration schema version. Currently always "v1". default: v1 nullable: true readOnly: true isEnabled: type: boolean description: >- Enables or disables noise suppression. When disabled, no noise suppression is applied regardless of other settings. default: false nullable: true type: allOf: - $ref: '#/components/schemas/NoiseSuppressionType' description: The noise suppression engine to use. default: none nullable: true noiseSuppressionLevel: type: integer description: >- The noise suppression level, from 0 (minimum) to 100 (maximum). Default is 75. Controls how aggressively background noise is filtered. format: int32 default: 75 nullable: true recordWriteStrategy: allOf: - $ref: '#/components/schemas/RecordWriteStrategy' description: |- Controls what audio is stored in the call recording. Default is "raw". default: raw nullable: true additionalProperties: false description: >- Noise suppression configuration for filtering background noise from call audio. Supports different noise suppression engines and controls recording strategy. ApiKeyAuthentication: allOf: - $ref: '#/components/schemas/McpAuthenticationBase' - required: - apiKey - type type: object properties: type: minLength: 1 enum: - apiKey type: string description: Authentication type. Always "apiKey" for API key authentication. readOnly: true apiKey: minLength: 1 type: string description: >- API key value to send with requests. This is the secret key provided by the MCP server. headerName: type: string description: >- HTTP header name for the API key. Common values include "X-API-Key", "Authorization", or custom header names required by the MCP server. nullable: true additionalProperties: false description: >- API key authentication for MCP server connections. Sends the API key in a custom HTTP header with each request to the MCP server. BearerAuthentication: allOf: - $ref: '#/components/schemas/McpAuthenticationBase' - required: - token - type type: object properties: type: minLength: 1 enum: - bearer type: string description: >- Authentication type. Always "bearer" for bearer token authentication. readOnly: true token: minLength: 1 type: string description: >- Bearer token value without the "Bearer " prefix. The system automatically adds the prefix when sending requests. This is the OAuth or JWT token provided by the MCP server or authorization service. additionalProperties: false description: >- Bearer token authentication for MCP server connections. The system automatically adds the "Bearer " prefix and sends the token in the Authorization header as "Bearer {token}" with each request to the MCP server. NoAuthentication: allOf: - $ref: '#/components/schemas/McpAuthenticationBase' - required: - type type: object properties: type: minLength: 1 enum: - none type: string description: Authentication type. Always "none" for no authentication. readOnly: true additionalProperties: false description: >- No authentication for MCP server connections. Use when the MCP server does not require authentication or authentication is handled through other means like network access controls. McpConnectionTransport: enum: - SSE - StreamableHTTP type: string description: >- Transport protocol used for Model Context Protocol (MCP) server communication. Determines how messages are exchanged between the agent and the MCP server. StartWebHookResponseAccept: type: object properties: accept: enum: - true type: boolean description: >- Accept indicator. Always true for this response type, indicating the call should proceed. default: true readOnly: true overrideAgentConfig: type: object description: >- Optional modified agent configuration to use for this call instead of the original. Allows customizing the agent's behavior, language model settings, prompts, voice, tools, or any other configuration based on external logic or data. The entire configuration is replaced, not merged, so include all settings you want to apply. default: null nullable: true additionalData: type: object additionalProperties: {} description: >- Additional data variables to use for this call. These values can be referenced in prompts using {{variableName}} syntax and override or supplement the agent's default AdditionalData. Useful for personalizing the conversation based on external context or user information. nullable: true overrideOutboundConfig: allOf: - $ref: '#/components/schemas/OutboundConfigOverride' description: >- Optional override for outbound SIP configuration. Allows changing the SIP credentials, server, or transport settings for this specific outbound call without modifying the agent's default SIP configuration. default: null nullable: true additionalProperties: false description: >- Response to accept a call and proceed with the conversation. Returned from the start webhook or used as a fallback when the start webhook fails. Allows optionally overriding the agent configuration, additional data variables, or outbound SIP settings for this specific call. StartWebHookResponseReject: required: - reasonMessage type: object properties: accept: enum: - false type: boolean description: >- Accept indicator. Always false for this response type, indicating the call should be rejected. default: false readOnly: true reasonMessage: minLength: 1 type: string description: >- Explanation of why the call is being rejected. This message is used for logging and debugging purposes to understand why external logic determined the call should not proceed. additionalProperties: false description: >- Response to reject a call and prevent the conversation from starting. Returned from the start webhook or used as a fallback when the start webhook fails. Causes the call to be declined with a reason message explaining why it was rejected. SpeedAdjustment: enum: - Disabled - OnRequest type: string description: >- Speech speed adjustment mode controlling whether the agent can change its speaking rate during conversations. Determines if users can request the agent to speak faster or slower to match their comprehension preferences. HttpMethod: enum: - GET - POST - PUT - PATCH - DELETE type: string description: >- HTTP method to use for webhook requests. Determines how the webhook data is sent to the target endpoint. POST is the most common choice for sending event data, while GET, PUT, PATCH, and DELETE can be used for REST API integrations that require specific HTTP verbs. BodyTextFormat: allOf: - $ref: '#/components/schemas/BodyFormat' - required: - template - type type: object properties: type: enum: - Text type: string description: Format type identifier. Always returns Text for this format. readOnly: true template: minLength: 1 type: string description: >- Text template defining the structure of the request body. Can include placeholders that are replaced with actual webhook payload values. The template is sent with Content-Type text/plain. additionalProperties: false description: >- Plain text body format using a template string. Allows embedding webhook payload values into a custom text template. Useful for integrating with systems that expect plain text payloads or custom string formats. BodyJSONFormat: allOf: - $ref: '#/components/schemas/BodyFormat' - required: - type type: object properties: type: enum: - Json type: string description: Format type identifier. Always returns Json for this format. readOnly: true template: type: array items: $ref: '#/components/schemas/BodyFormatValue' description: >- List of field mappings defining which values to extract from the webhook payload and where to place them in the output JSON object. Each mapping specifies a JSON path in the payload and where to put the value in the result. nullable: true additionalProperties: false description: >- JSON body format using field mappings from the webhook payload. Constructs a JSON object by extracting values from the payload using JSON path expressions. Sends the result with Content-Type application/json. The most common format for modern API integrations. BodyFormFormat: allOf: - $ref: '#/components/schemas/BodyFormat' - required: - type type: object properties: type: enum: - Form type: string description: Format type identifier. Always returns Form for this format. readOnly: true template: type: array items: $ref: '#/components/schemas/BodyFormatValue' description: >- List of field mappings defining which values to extract from the webhook payload and how to encode them as form fields. Each mapping specifies a JSON path in the payload and the corresponding form field name. nullable: true additionalProperties: false description: >- Form-encoded body format using field mappings from the webhook payload. Constructs URL-encoded form data by extracting values from the payload using JSON path expressions. Sends the result with Content-Type application/x-www-form-urlencoded. Used for integrating with systems that expect traditional HTML form submissions. TransferFeatureBase: required: - description - type type: object properties: type: allOf: - $ref: '#/components/schemas/TransferType' description: >- Transfer type discriminator indicating which transfer mechanism is used. readOnly: true version: type: string description: Configuration schema version. Currently always "v1". nullable: true readOnly: true description: minLength: 1 type: string description: >- Description of when and why this transfer should be used. Guides the language model in deciding when to invoke this transfer option during conversations. Should explain the circumstances or user requests that trigger this transfer. failoverBehavior: allOf: - $ref: '#/components/schemas/FailoverBehavior' description: >- Behavior when the transfer fails due to the destination being unavailable, busy, or unreachable. Controls whether the conversation continues with the AI agent or ends, and what message to communicate to the user about the failure. nullable: true isEnabled: type: boolean description: >- Enables or disables this transfer option. When disabled, the language model cannot select this transfer even if the description matches the conversation context. nullable: true additionalProperties: false description: >- Base class for call transfer configurations. Defines common properties for all transfer types including description, failover behavior when transfers fail, and enable/disable control. Concrete implementations include cold transfers for immediate handoff, warm transfers with agent introduction, and HTTP transfers for dynamic routing decisions. TransferRoute: required: - description - endpointDestinations - name type: object properties: name: maxLength: 100 minLength: 1 pattern: ^[a-zA-Z\-_][a-zA-Z0-9\-_]*$ type: string description: >- Route name identifier. Must start with a letter, hyphen, or underscore and contain only letters, numbers, hyphens, and underscores. Used by the language model to reference this route when making transfer decisions. endpointDestinations: minItems: 1 type: array items: $ref: '#/components/schemas/EndpointDescription' description: >- List of destination endpoints for this route. Can include multiple endpoints for failover or load balancing. The system attempts to transfer to these endpoints in order until one succeeds. description: maxLength: 500 minLength: 1 type: string description: >- Description of when this route should be used. Guides the language model in selecting the appropriate route based on conversation context, user requests, or caller needs. Should explain what type of assistance or department this route connects to. additionalProperties: false description: >- Named transfer routing option for intelligent destination selection. Defines a named route with one or more endpoint destinations that the language model can choose based on conversation context. Enables dynamic routing to different departments, specialists, or services by allowing the model to select the most appropriate route based on the user's needs described in the route description. HoldMediaConfig: type: object properties: isEnabled: type: boolean description: >- Enables or disables hold media playback during transfers. When enabled, the specified media file plays on loop until the transfer completes. When disabled or not specified, the agent uses its voice to communicate with users during transfer. default: false nullable: true mediaFilename: type: string description: >- Filename of the uploaded media file to play as hold music. Should be the filename of a media file that has been uploaded to the system. The file plays on loop during the transfer process. default: null nullable: true additionalProperties: false description: >- Hold media configuration for playing audio files during call transfers. When enabled, plays the specified media file on loop while the transfer is in progress instead of using the agent's voice. Provides professional hold music or informational messages while users wait for the transfer to complete. The media file must be uploaded to the system before being referenced here. OptionalOutboundSipConfig: type: object properties: domain: type: string description: >- SIP domain for authentication realm as defined in RFC 3261. Used in SIP authentication challenges. default: null nullable: true password: type: string description: >- Password for SIP authentication as defined in RFC 3261. Used for authenticating outbound calls. default: null nullable: true fromUser: type: string description: >- From header user portion for outbound calls. Determines what user identifier appears in the SIP From header when making calls. default: null nullable: true transport: allOf: - $ref: '#/components/schemas/TransportProtocol' description: >- Transport protocol for SIP communication. Supported values are UDP (default), TCP, or TLS for encrypted connections. default: udp nullable: true cpsLimit: type: integer description: >- Calls per second limit for rate limiting outbound calls. Prevents exceeding provider rate limits. format: int32 default: null nullable: true cpsKey: type: string description: >- Calls per second rate limiting key. Used to group multiple configurations under the same rate limit when specified. default: null nullable: true displayName: type: string description: >- Display name for caller ID. Appears in the From header display name when making outbound calls. default: null nullable: true type: enum: - v1 type: string description: >- Configuration version type. Always "v1" for inline credentials configuration. default: v1 nullable: true readOnly: true server: type: string description: >- SIP server hostname or IP address with optional port as defined in RFC 3261. Optional in this variant. nullable: true authUser: type: string description: >- Authentication user identifier as defined in RFC 3261. Optional in this variant. nullable: true additionalProperties: false description: >- Optional variant of outbound SIP configuration V1 where Server and AuthUser are not required. Used in contexts where SIP configuration may be partially specified or inherited from defaults. SmartV1Behavior: allOf: - $ref: '#/components/schemas/WarmTransferBehaviorBase' - required: - type type: object properties: type: enum: - smartV1 type: string description: >- Warm transfer behavior type discriminator determining how the agent interacts with operators before bridging calls. Static behavior uses predefined rules with explicit wait conditions like DTMF tones, audio frequencies, or voice detection to determine when the operator is ready. SmartV1 behavior uses AI to intelligently detect operator readiness and adapt conversation flow dynamically. Unspecified serves as fallback for forward compatibility with future behavior types. readOnly: true additionalInstructions: type: string description: >- Optional instructions to customize how the AI agent interacts with the operator before bridging the call. Provides context-specific guidance for different scenarios, such as explaining the caller's situation, verifying operator availability, or following specific handoff protocols. Helps ensure smooth operator interactions while maintaining flexibility for different business processes. default: null nullable: true additionalProperties: false description: >- AI-driven warm transfer behavior that intelligently detects when the human operator is ready and adapts the conversation flow dynamically. Uses language model intelligence to determine optimal timing for bridging the call based on the operator's voice responses and context, rather than relying on predefined wait conditions. Optional additional instructions can customize how the AI interacts with the operator before completing the transfer, allowing context-specific guidance for different operator scenarios or business processes. StaticBehavior: allOf: - $ref: '#/components/schemas/WarmTransferBehaviorBase' - required: - type type: object properties: type: enum: - static type: string description: >- Warm transfer behavior type discriminator determining how the agent interacts with operators before bridging calls. Static behavior uses predefined rules with explicit wait conditions like DTMF tones, audio frequencies, or voice detection to determine when the operator is ready. SmartV1 behavior uses AI to intelligently detect operator readiness and adapt conversation flow dynamically. Unspecified serves as fallback for forward compatibility with future behavior types. readOnly: true waitFirstAnswerFromOperator: type: boolean description: >- Wait for voice detection from the operator before bridging the call. When enabled, the system waits for the operator to start speaking before connecting the caller, ensuring the operator is actively engaged and ready. When disabled, the call bridges immediately once other configured wait conditions are met or the timeout expires. default: false nullable: true waitToneFromOperator: type: array items: type: number format: double description: >- Audio frequency tones to wait for from the operator before bridging, specified in Hertz. Used to detect specific audio signals from the operator's phone system or IVR, such as 400Hz for readiness signals. The system listens for any of the specified frequencies and bridges the call when detected. Useful for integrating with legacy phone systems or custom operator signaling protocols. default: null nullable: true waitDTMFFromOperator: type: array items: type: string description: >- DTMF key press tones to wait for from the operator before bridging the call. Accepts digit keys 0-9 and special keys like pound or star. The system waits for the operator to press any of the specified keys as an acknowledgment signal before connecting the caller. Common in call center environments where operators confirm readiness by pressing a specific key. default: null nullable: true waitChecksTimeoutInS: type: number description: >- Maximum time in seconds to wait for operator readiness signals before timing out. If the operator does not provide the expected voice, DTMF, or frequency tone signals within this timeout, the wait conditions are bypassed and the call proceeds according to failover configuration. Prevents indefinite waiting when operators are unavailable or unresponsive. format: double default: 4 nullable: true sayPhraseToOperator: allOf: - $ref: '#/components/schemas/SayPhraseToOperator' description: >- Phrase configuration for introducing the caller to the operator before bridging. Allows the AI agent to provide context about the caller's needs, reason for transfer, or conversation summary to help the operator assist effectively. Can be configured to speak before or after the operator acknowledges, depending on call center workflow preferences. default: null nullable: true additionalProperties: false description: >- Static warm transfer behavior with explicit wait conditions before bridging the caller to the operator. Allows precise control over transfer timing by waiting for specific operator readiness signals including voice detection, DTMF key presses, or audio frequency tones. Supports configurable timeout for wait conditions and optional phrase to introduce the caller to the operator before bridging. Useful for structured call center environments with standardized operator protocols or IVR systems requiring specific acknowledgment signals. CallerIdMode: enum: - userPhoneNumber - agentPhoneNumber type: string description: >- Caller ID display mode for warm transfers. Determines which phone number appears on the operator's caller ID when the AI agent calls them during a warm transfer. Controls whether the operator sees the original user's number or the agent's number, affecting how operators identify and prioritize incoming transfer calls. ThreewaySmartV1Behavior: allOf: - $ref: '#/components/schemas/ThreewayMessagingBehaviorBase' - required: - type type: object properties: type: enum: - smartV1 type: string description: >- Threeway messaging behavior type discriminator determining how announcement phrases are delivered to both caller and operator simultaneously during warm transfer bridging. Static behavior uses predefined phrases spoken unchanged to both parties for consistent messaging. SmartV1 behavior uses AI to generate contextual announcement phrases based on conversation history and optional instructions, adapting to each transfer scenario. Unspecified serves as fallback for forward compatibility with future behavior types. The phrase is delivered to both the parent conversation channel and warm transfer child channel simultaneously right before connecting the caller to the operator. readOnly: true additionalInstructions: type: string description: >- Optional instructions to customize how the AI generates the announcement phrase spoken to both parties simultaneously. Guides the language model in crafting appropriate handoff messaging, such as maintaining professional tone, including issue summaries, providing reassurance, or mentioning specialist involvement. When not specified, the phrase is generated based solely on conversation context without additional constraints. default: null nullable: true additionalProperties: false description: >- AI-driven threeway messaging behavior that uses language model intelligence to generate contextual announcement phrases spoken to both caller and operator simultaneously during warm transfer bridging. The phrase is dynamically generated based on conversation transcript history and call context, adapting the handoff announcement to each unique transfer scenario. Optional additional instructions can guide the phrase generation to maintain specific tone, include issue summaries, emphasize reassurance, or mention specialist involvement. When additional instructions are not provided, the phrase is generated solely from conversation context. The unified announcement is delivered to both channels simultaneously right before bridging the call, ensuring consistent messaging for both parties during the transition. ThreewayStaticBehavior: allOf: - $ref: '#/components/schemas/ThreewayMessagingBehaviorBase' - required: - type type: object properties: type: enum: - static type: string description: >- Threeway messaging behavior type discriminator determining how announcement phrases are delivered to both caller and operator simultaneously during warm transfer bridging. Static behavior uses predefined phrases spoken unchanged to both parties for consistent messaging. SmartV1 behavior uses AI to generate contextual announcement phrases based on conversation history and optional instructions, adapting to each transfer scenario. Unspecified serves as fallback for forward compatibility with future behavior types. The phrase is delivered to both the parent conversation channel and warm transfer child channel simultaneously right before connecting the caller to the operator. readOnly: true phrase: type: string description: >- Phrase spoken to both caller and operator simultaneously right before bridging the call. Should be clear and concise as both parties will hear it at the same moment during the transfer transition. Common examples include connection announcements like "Connecting you now", hold notifications like "Please hold while I connect you", or representative introduction messages like "You are now being connected to a representative". default: null nullable: true additionalProperties: false description: >- Static threeway messaging behavior with a predefined phrase spoken to both caller and operator simultaneously during warm transfer bridging. The configured phrase is delivered unchanged to both parties, providing consistent and predictable messaging during all transfers. Useful for maintaining standardized handoff messaging like connection announcements or hold notifications that both the caller and operator need to hear at the transition moment. PostCallAnalysisLabel: required: - description - name - type type: object properties: name: maxLength: 100 minLength: 1 pattern: ^[a-zA-Z\-_][a-zA-Z0-9\-_]*$ type: string description: >- Field name identifier. Must start with a letter, hyphen, or underscore and contain only letters, numbers, hyphens, and underscores. This name is used as the key in the analysis results. description: maxLength: 500 minLength: 1 type: string description: >- Description of what information to extract and how to interpret it. Guides the language model in identifying and extracting the relevant data from the conversation. Should clearly explain what to look for and how to categorize or format the extracted information. type: allOf: - $ref: '#/components/schemas/PostCallAnalysisLabelType' description: >- Data type for the extracted value. Determines the structure and format of the extracted information and constrains what the language model can return. values: type: array items: type: string description: >- Allowed values for enum type labels. When Type is Enum, the extracted value must be one of these predefined options. The language model selects the most appropriate value from this list based on the conversation content. nullable: true required: type: boolean description: >- Indicates whether this field must be populated. When true, the analysis will always attempt to extract a value for this field. When false, the field may be left empty if the relevant information is not found in the conversation. nullable: true additionalProperties: false description: >- Data field definition for post-call analysis extraction. Specifies a piece of information to identify and extract from conversation transcripts such as customer sentiment, appointment details, or business outcomes. The language model analyzes the completed conversation and attempts to populate this field based on the description and type constraints. SmartV1IvrDetectionBehavior: allOf: - $ref: '#/components/schemas/IvrDetectionBehaviorBase' - required: - type type: object properties: type: enum: - smartV1 type: string description: >- Behavior type identifier. Always returns SmartV1 for this AI-generated approach. readOnly: true additionalInstructions: type: string description: >- Optional instructions to guide the language model in generating voicemail messages. Provides context about what information to include, tone to use, or specific details to mention. Supports variable interpolation using {{variableName}} syntax for dynamic instructions based on call data. default: null nullable: true additionalProperties: false description: >- Smart version 1 IVR detection behavior using language model generation for contextual voicemail messages. When an answering machine is detected, the agent uses AI to generate an appropriate message based on the conversation purpose and any additional instructions. Creates more natural and context-aware voicemail messages than static phrases. StaticIvrDetectionBehavior: allOf: - $ref: '#/components/schemas/IvrDetectionBehaviorBase' - required: - type type: object properties: type: enum: - static type: string description: >- Behavior type identifier. Always returns Static for this approach. readOnly: true staticPhrase: type: string description: >- Voicemail message to speak when an answering machine is detected. The agent says this exact phrase after detection. Supports variable interpolation using {{variableName}} syntax to include dynamic data like caller names, phone numbers, or appointment details in the message. default: null nullable: true additionalProperties: false description: >- Static IVR detection behavior using a predefined voicemail message. When an answering machine is detected, the agent says the exact configured phrase and ends the call. Supports variable interpolation using {{variableName}} syntax for personalizing messages with call-specific data. KBLink: required: - id type: object properties: id: maxLength: 255 minLength: 1 type: string description: >- Unique identifier of the knowledge base to link. References a knowledge base that has been created and contains indexed documents, articles, or reference materials that the agent can search. additionalProperties: false description: >- Knowledge base link reference for RAG (Retrieval-Augmented Generation) feature. Points to a knowledge base that the agent can search during conversations to retrieve relevant information for answering questions. The knowledge base must be created and populated with documents before being linked to the agent. FillerStaticStrategy: allOf: - $ref: '#/components/schemas/FillerStrategy' - required: - type type: object properties: type: enum: - static type: string description: >- Strategy type identifier. Always returns Static for this filler approach. readOnly: true texts: type: array items: type: string description: >- List of filler phrases the agent can use. The agent randomly selects from these options when inserting verbal pauses. Common examples include "um", "uh", "let me see", "well", "you know", "hmm", or "let me think". Customize this list to control the agent's hesitation patterns and speaking style. default: - um nullable: true additionalProperties: false description: >- Static filler word strategy using a predefined list of verbal pauses and hesitation sounds. The agent randomly selects from the configured phrases when inserting fillers during conversations. Allows customizing which specific filler words the agent uses to match desired speaking style and naturalness level. SmartV1BackchannelBehavior: allOf: - $ref: '#/components/schemas/BackchannelBehaviorBase' - required: - type type: object properties: type: enum: - smartV1 type: string description: >- Behavior type identifier. Always returns SmartV1 for this AI-driven approach. readOnly: true additionalInstructions: type: string description: >- Optional instructions to guide the AI in producing backchannels. Provides context about when or how to acknowledge, tone to use, or specific phrases to prefer. Supports variable interpolation using {{variableName}} syntax for dynamic instructions based on conversation data. default: null nullable: true frequency: type: number description: >- Frequency of backchannel insertions ranging from 0.0 to 1.0. Lower values make backchannels very rare, 0.5 provides moderate acknowledgments, and higher values produce frequent listening cues. Controls how often the AI decides to signal active listening during user speech. format: double default: 0.5 nullable: true additionalProperties: false description: >- Smart version 1 backchannel behavior using AI to determine natural moments for acknowledgments. The agent uses intelligent detection to identify appropriate conversation points for inserting listening cues based on context, user pauses, and conversation flow. Creates more natural and contextually appropriate backchannels than static timing rules. StaticAdvancedBackchannelBehavior: allOf: - $ref: '#/components/schemas/BackchannelBehaviorBase' - required: - type type: object properties: type: enum: - staticAdvanced type: string description: >- Behavior type identifier. Always returns StaticAdvanced for this approach. readOnly: true phrases: type: array items: type: string description: >- List of acknowledgment phrases the agent can use for backchannels. The agent randomly selects from these options when timing conditions are met. Common examples include "uh-huh", "mm-hmm", "I see", "right", "okay", or "go on". Supports variable interpolation using {{variableName}} syntax for dynamic phrases. default: null nullable: true minLengthVoice: type: number description: >- Minimum duration in seconds the user must speak before a backchannel can be triggered. Prevents acknowledgments from interrupting very short utterances. The agent waits at least this long into user speech before considering backchannel insertion. format: double default: 10 nullable: true minLengthVoiceDeviation: type: number description: >- Percentage deviation for minimum voice length randomization, ranging from 0.0 to 1.0. Multiplied by MinLengthVoice to introduce natural variation in when backchannels can occur. Higher values create more unpredictability in timing for a more human-like pattern. format: double default: 0.25 nullable: true frequency: type: number description: >- Frequency of backchannel insertions ranging from 0.0 to 1.0. Controls how often acknowledgments occur when timing conditions are met. Lower values make backchannels rarer, higher values produce more frequent listening cues. format: double default: 0.5 nullable: true minCooldown: type: number description: >- Minimum time in seconds that must elapse between consecutive backchannels. Prevents acknowledgments from occurring too close together. The agent waits at least this long after one backchannel before producing another. format: double default: 8 nullable: true minCooldownDeviation: type: number description: >- Percentage deviation for cooldown randomization, ranging from 0.0 to 1.0. Multiplied by MinCooldown to introduce natural variation in spacing between backchannels. Higher values create more unpredictable timing for a more human-like rhythm. format: double default: 0.25 nullable: true additionalProperties: false description: >- Advanced static backchannel behavior with fine-tuned timing controls for natural acknowledgment patterns. Provides detailed configuration for minimum voice duration before triggering backchannels, cooldown periods between acknowledgments, and deviation parameters for randomization. Creates more sophisticated listening behavior than basic static mode by controlling timing thresholds and introducing natural variation. StaticBackchannelBehavior: allOf: - $ref: '#/components/schemas/BackchannelBehaviorBase' - required: - type type: object properties: type: enum: - static type: string description: >- Behavior type identifier. Always returns Static for this approach. readOnly: true phrases: type: array items: type: string description: >- List of acknowledgment phrases the agent can use for backchannels. The agent randomly selects from these options when inserting listening cues. Common examples include "uh-huh", "mm-hmm", "I see", "right", "okay", or "go on". Supports variable interpolation using {{variableName}} syntax for dynamic phrases. default: null nullable: true frequency: type: number description: >- Frequency of backchannel insertions ranging from 0.0 to 1.0. Lower values make backchannels very rare, 0.5 provides moderate acknowledgments, and higher values produce frequent listening cues. Controls how often the agent signals active listening during user speech. format: double default: 0.5 nullable: true additionalProperties: false description: >- Static backchannel behavior using predefined acknowledgment phrases with basic frequency control. The agent randomly selects from the configured list of listening cues and inserts them while the user speaks. Provides simple configuration for adding natural verbal feedback without complex timing parameters. TurnTakingVad: enum: - asap_v1 - asap_v2 type: string description: >- Voice activity detection mode used to decide when the caller's turn has ended. Values map directly to the DashaScript connect option named "vad". NoiseSuppressionType: enum: - none - krisp type: string description: The noise suppression engine type. RecordWriteStrategy: enum: - raw - cleaned type: string description: Controls what audio is stored in the call recording. McpAuthenticationBase: type: object properties: type: type: string description: >- Authentication type identifier. Supported values include "none", "apiKey", "bearer", and custom types. nullable: true readOnly: true additionalProperties: false description: >- Base class for Model Context Protocol authentication methods. Supports multiple authentication types including none, API key, bearer token, and custom authentication schemes. OutboundConfigOverride: type: object properties: phoneNumber: maxLength: 255 type: string description: >- Phone number to use for this call instead of the agent's default. Overrides which number appears as the caller ID and which SIP registration is used for the call. default: null nullable: true credentialsId: type: string description: >- SIP credentials to use for this call instead of the agent's default. Allows routing the call through a different SIP trunk, useful for failover, cost optimization, or geographic routing. format: uuid default: null nullable: true displayName: maxLength: 255 type: string description: >- Display name to use for caller ID for this call instead of the phone number's default. Overrides the caller ID name shown to the recipient. default: null nullable: true additionalProperties: false description: >- Override outbound SIP configuration for a specific call. Allows using different phone number, credentials, or caller ID for individual calls without changing the agent's default configuration. Useful for multi-tenant scenarios, A/B testing different trunks, or presenting different caller IDs per customer. BodyFormat: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/BodyFormatEnum' description: >- Format type discriminator indicating which body format implementation is used. Determines whether the payload is sent as text, JSON, or form data. readOnly: true additionalProperties: false description: >- Base class for webhook request body format configurations. Defines how the webhook payload is encoded and structured. Concrete implementations include text templates, JSON object mapping, and form-encoded data. BodyFormatValue: required: - jsonPath type: object properties: jsonPath: minLength: 1 type: string description: >- JSON path in the output where the extracted value should be placed. For JSON format, this defines the structure of the output object. For form format, this is the form field name. argumentJsonPath: type: string description: >- JSON path in the webhook payload to extract the value from. When specified, the value at this path in the source payload is copied to the JsonPath location in the output. If not specified, the DefaultValue is used instead. nullable: true defaultValue: description: >- Default value to use when ArgumentJsonPath is not specified or when the path doesn't exist in the payload. Allows providing fallback values to ensure the output always has required fields populated. nullable: true additionalProperties: false description: >- Field mapping configuration for extracting values from webhook payloads. Defines how to extract a value from the source payload using JSON path and where to place it in the output, with support for default values when the source path doesn't exist. TransferType: enum: - unspecified - cold - warm - http type: string description: >- Call transfer type determining how the call is handed off to a human operator or different destination. Different transfer types control the handoff mechanism, agent involvement during transfer, and whether decisions are made dynamically via HTTP requests. FailoverBehavior: type: object properties: continueConversation: type: boolean description: >- Controls whether the AI agent continues the conversation after the transfer fails. When true, the agent resumes talking to the user and can attempt alternative solutions. When false, the call ends after explaining the transfer failure. default: false nullable: true systemMessage: type: string description: >- System message added to the conversation context when the transfer fails. Provides internal context to the language model about the failure, influencing how it responds to the user after the failover. Not spoken aloud but affects the agent's behavior and understanding. default: null nullable: true staticPhrase: type: string description: >- Phrase spoken to the user when the transfer fails. Explains that the operator is unavailable and what happens next. Should provide clear communication about the failure and set expectations for the user. default: null nullable: true additionalProperties: false description: >- Failover behavior configuration controlling what happens when a transfer fails. Defines whether the conversation continues with the AI agent or ends, and what messages are communicated to the user about the failed transfer. Allows graceful handling of unreachable operators, busy signals, or network failures during transfer attempts. EndpointDescription: required: - endpoint type: object properties: endpoint: minLength: 1 type: string description: >- Destination phone number or SIP address. Supports E.164 phone numbers like "+1234567890" or SIP URIs like "sip:operator@example.com". The transfer system attempts to connect to this endpoint when this destination is selected. additionalProperties: false description: >- Transfer endpoint destination specification. Defines a phone number or SIP address that calls can be transferred to. Used in transfer routes to specify possible destinations with support for failover when multiple endpoints are configured. WarmTransferBehaviorBase: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/WarmTransferBehavior' description: >- Warm transfer behavior type discriminator determining how the agent interacts with operators before bridging calls. Static behavior uses predefined rules with explicit wait conditions like DTMF tones, audio frequencies, or voice detection to determine when the operator is ready. SmartV1 behavior uses AI to intelligently detect operator readiness and adapt conversation flow dynamically. Unspecified serves as fallback for forward compatibility with future behavior types. readOnly: true additionalProperties: false description: >- Base class for warm transfer behavior configurations controlling how the AI agent interacts with the human operator before bridging the caller. Defines whether to use static rules with explicit wait conditions and operator greetings, or smart AI-driven behavior that adapts to the operator's responses. Determines timing of call bridging based on operator availability signals like voice detection, DTMF tones, or specific audio frequencies. SayPhraseToOperator: required: - phrase type: object properties: beforeOperator: type: boolean description: >- Controls timing of when the phrase is spoken. When true, the agent speaks immediately when the call connects, potentially before the operator fully answers. When false, waits for the operator to acknowledge before speaking the phrase. default: false nullable: true phrase: minLength: 1 type: string description: >- Phrase to speak to the operator when they answer. Should provide relevant context about the caller or their needs. Can include conversation summaries, caller identification, or reason for transfer to help the operator assist the user effectively. additionalProperties: false description: >- Configuration for speaking a phrase to the operator during warm transfer. Defines what the AI agent says to the human operator when they answer, and whether to speak before or after they pick up. Allows providing context, caller information, or conversation summaries to help the operator understand the caller's needs before being connected. ThreewayMessagingBehaviorBase: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/ThreewayMessagingBehavior' description: >- Threeway messaging behavior type discriminator determining how announcement phrases are delivered to both caller and operator simultaneously during warm transfer bridging. Static behavior uses predefined phrases spoken unchanged to both parties for consistent messaging. SmartV1 behavior uses AI to generate contextual announcement phrases based on conversation history and optional instructions, adapting to each transfer scenario. Unspecified serves as fallback for forward compatibility with future behavior types. The phrase is delivered to both the parent conversation channel and warm transfer child channel simultaneously right before connecting the caller to the operator. readOnly: true additionalProperties: false description: >- Base class for threeway messaging behavior configurations controlling how announcement phrases are delivered to both caller and operator simultaneously during warm transfer bridging. Defines whether to use static predefined phrases for consistent messaging, or AI-driven smart behavior that generates contextual announcements based on conversation history. The unified phrase is spoken to both parties at the same moment right before connecting them, ensuring both the caller and operator hear the same transition message as the call is bridged. PostCallAnalysisLabelType: enum: - string - number - boolean - enum type: string description: >- Data type for post-call analysis label values. Determines how the extracted information is structured and what kind of data the language model should extract from the conversation transcript. IvrDetectionBehaviorBase: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/IvrDetectionBehavior' description: >- Behavior type discriminator indicating which message generation approach is used. readOnly: true additionalProperties: false description: >- Base class for IVR and answering machine detection behaviors. Defines how the agent responds when calls are answered by voicemail or automated systems. Concrete implementations include static behaviors with predefined messages and smart behaviors that generate contextual responses using language models. FillerStrategy: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/FillerStrategyEnum' description: >- Strategy type discriminator indicating which filler selection approach is used. readOnly: true additionalProperties: false description: >- Base class for filler word selection strategies. Defines how the agent chooses and inserts verbal pauses and hesitation sounds to make speech more natural. Concrete implementations include static strategies with predefined filler lists. BackchannelBehaviorBase: required: - type type: object properties: type: allOf: - $ref: '#/components/schemas/BackchannelBehavior' description: >- Behavior type discriminator indicating which backchannel approach is used. readOnly: true additionalProperties: false description: >- Base class for backchannel acknowledgment behaviors. Defines how the agent produces verbal listening cues while users speak. Concrete implementations include static behaviors with predefined phrases and smart behaviors that use AI to determine natural moments for acknowledgments. BodyFormatEnum: enum: - Text - Json - Form type: string description: >- Content format for the webhook request body. Determines how the webhook payload is encoded and what Content-Type header is sent. Different formats are appropriate for different receiving endpoints and integration requirements. WarmTransferBehavior: enum: - unspecified - static - smartV1 type: string description: >- Warm transfer behavior type discriminator determining how the agent interacts with operators before bridging calls. Static behavior uses predefined rules with explicit wait conditions like DTMF tones, audio frequencies, or voice detection to determine when the operator is ready. SmartV1 behavior uses AI to intelligently detect operator readiness and adapt conversation flow dynamically. Unspecified serves as fallback for forward compatibility with future behavior types. ThreewayMessagingBehavior: enum: - Unspecified - static - smartV1 type: string description: >- Threeway messaging behavior type discriminator determining how announcement phrases are delivered to both caller and operator simultaneously during warm transfer bridging. Static behavior uses predefined phrases spoken unchanged to both parties for consistent messaging. SmartV1 behavior uses AI to generate contextual announcement phrases based on conversation history and optional instructions, adapting to each transfer scenario. Unspecified serves as fallback for forward compatibility with future behavior types. The phrase is delivered to both the parent conversation channel and warm transfer child channel simultaneously right before connecting the caller to the operator. IvrDetectionBehavior: enum: - unspecified - static - smartV1 type: string description: >- Behavior type for handling answering machine and IVR detection. Determines how the agent responds when a call is answered by voicemail or an automated system. Different behaviors control message generation and delivery strategies. FillerStrategyEnum: enum: - unspecified - static type: string description: >- Filler word selection strategy type. Determines how the agent chooses which filler words to use and when to insert them during conversations to sound more natural and human-like. BackchannelBehavior: enum: - unspecified - static - staticAdvanced - smartV1 type: string description: >- Backchannel acknowledgment behavior type. Determines how the agent produces verbal listening cues while the user speaks. Different behaviors control phrase selection, timing, and frequency of acknowledgments like "uh-huh", "mm-hmm", or "I see" to signal active listening. ````