telnyx 5.91.0 → 5.93.0

data/rbi/telnyx/models/ai/inference_embedding.rbi

@@ -20,9 +20,11 @@ module Telnyx
         sig { returns(String) }
         attr_accessor :instructions
 
-        # ID of the model to use. You can use the
+        # ID of the model to use when `external_llm` is not set. You can use the
         # [Get models API](https://developers.telnyx.com/api-reference/chat/get-available-models)
-        # to see all of your available models,
+        # to see available models. If `external_llm` is provided, the assistant uses
+        # `external_llm` instead of this field. If neither `model` nor `external_llm` is
+        # provided, Telnyx applies the default model.
         sig { returns(String) }
         attr_accessor :model
 
@@ -42,10 +44,24 @@ module Telnyx
         sig { params(dynamic_variables: T::Hash[Symbol, T.anything]).void }
         attr_writer :dynamic_variables
 
-        # If the dynamic_variables_webhook_url is set for the assistant, we will send a
-        # request at the start of the conversation. See our
-        # [guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
-        # for more information.
+        # Timeout in milliseconds for the dynamic variables webhook. Must be between 1 and
+        # 10000 ms. If the webhook does not respond within this timeout, the call proceeds
+        # with default values. See the
+        # [dynamic variables guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables).
+        sig { returns(T.nilable(Integer)) }
+        attr_reader :dynamic_variables_webhook_timeout_ms
+
+        sig { params(dynamic_variables_webhook_timeout_ms: Integer).void }
+        attr_writer :dynamic_variables_webhook_timeout_ms
+
+        # If `dynamic_variables_webhook_url` is set, Telnyx sends a POST request to this
+        # URL at the start of the conversation to resolve dynamic variables. **Gotcha:**
+        # the webhook response must wrap variables under a top-level `dynamic_variables`
+        # object, e.g. `{"dynamic_variables": {"customer_name": "Jane"}}`. Returning a
+        # flat object will be ignored and variables will fall back to their defaults. See
+        # the
+        # [dynamic variables guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
+        # for the full request/response format and timeout behavior.
         sig { returns(T.nilable(String)) }
         attr_reader :dynamic_variables_webhook_url
 
@@ -66,16 +82,27 @@ module Telnyx
         end
         attr_writer :enabled_features
 
-        sig { returns(T.nilable(Telnyx::AI::ExternalLlm)) }
+        sig { returns(T.nilable(Telnyx::AI::InferenceEmbedding::ExternalLlm)) }
         attr_reader :external_llm
 
-        sig { params(external_llm: Telnyx::AI::ExternalLlm::OrHash).void }
+        sig do
+          params(
+            external_llm: Telnyx::AI::InferenceEmbedding::ExternalLlm::OrHash
+          ).void
+        end
         attr_writer :external_llm
 
-        sig { returns(T.nilable(Telnyx::AI::FallbackConfig)) }
+        sig do
+          returns(T.nilable(Telnyx::AI::InferenceEmbedding::FallbackConfig))
+        end
         attr_reader :fallback_config
 
-        sig { params(fallback_config: Telnyx::AI::FallbackConfig::OrHash).void }
+        sig do
+          params(
+            fallback_config:
+              Telnyx::AI::InferenceEmbedding::FallbackConfig::OrHash
+          ).void
+        end
         attr_writer :fallback_config
 
         # Text that the assistant will use to start the conversation. This may be
@@ -104,17 +131,75 @@ module Telnyx
         end
         attr_writer :insight_settings
 
-        # This is only needed when using third-party inference providers. The `identifier`
-        # for an integration secret
+        # Connected integrations attached to the assistant. The catalog of available
+        # integrations is at `/ai/integrations`; the user's connected integrations are at
+        # `/ai/integrations/connections`. Each item references a catalog integration by
+        # `integration_id`.
+        sig do
+          returns(
+            T.nilable(T::Array[Telnyx::AI::InferenceEmbedding::Integration])
+          )
+        end
+        attr_reader :integrations
+
+        sig do
+          params(
+            integrations:
+              T::Array[Telnyx::AI::InferenceEmbedding::Integration::OrHash]
+          ).void
+        end
+        attr_writer :integrations
+
+        # Settings for interruptions and how the assistant decides the user has finished
+        # speaking. These timings are most relevant when using non turn-taking
+        # transcription models. For turn-taking models like `deepgram/flux`, end-of-turn
+        # behavior is controlled by the transcription end-of-turn settings under
+        # `transcription.settings` (`eot_threshold`, `eot_timeout_ms`,
+        # `eager_eot_threshold`).
+        sig do
+          returns(
+            T.nilable(Telnyx::AI::InferenceEmbedding::InterruptionSettings)
+          )
+        end
+        attr_reader :interruption_settings
+
+        sig do
+          params(
+            interruption_settings:
+              Telnyx::AI::InferenceEmbedding::InterruptionSettings::OrHash
+          ).void
+        end
+        attr_writer :interruption_settings
+
+        # This is only needed when using third-party inference providers selected by
+        # `model`. The `identifier` for an integration secret
         # [/v2/integration_secrets](https://developers.telnyx.com/api-reference/integration-secrets/create-a-secret)
-        # that refers to your LLM provider's API key. Warning: Free plans are unlikely to
-        # work with this integration.
+        # that refers to your LLM provider's API key. For bring-your-own endpoint
+        # authentication, use `external_llm.llm_api_key_ref` instead. Warning: Free plans
+        # are unlikely to work with this integration.
         sig { returns(T.nilable(String)) }
         attr_reader :llm_api_key_ref
 
         sig { params(llm_api_key_ref: String).void }
         attr_writer :llm_api_key_ref
 
+        # MCP servers attached to the assistant. Create MCP servers with
+        # `/ai/mcp_servers`, then reference them by `id` here.
+        sig do
+          returns(
+            T.nilable(T::Array[Telnyx::AI::InferenceEmbedding::McpServer])
+          )
+        end
+        attr_reader :mcp_servers
+
+        sig do
+          params(
+            mcp_servers:
+              T::Array[Telnyx::AI::InferenceEmbedding::McpServer::OrHash]
+          ).void
+        end
+        attr_writer :mcp_servers
+
         sig { returns(T.nilable(Telnyx::AI::MessagingSettings)) }
         attr_reader :messaging_settings
 
@@ -137,13 +222,17 @@ module Telnyx
         # can execute multiple parallel or sequential tools during this phase.
         # Telephony-control tools (e.g. hangup, transfer) are unavailable
         # post-conversation. Beta feature.
-        sig { returns(T.nilable(Telnyx::AI::PostConversationSettings)) }
+        sig do
+          returns(
+            T.nilable(Telnyx::AI::InferenceEmbedding::PostConversationSettings)
+          )
+        end
         attr_reader :post_conversation_settings
 
         sig do
           params(
             post_conversation_settings:
-              Telnyx::AI::PostConversationSettings::OrHash
+              Telnyx::AI::InferenceEmbedding::PostConversationSettings::OrHash
           ).void
         end
         attr_writer :post_conversation_settings
@@ -156,6 +245,21 @@ module Telnyx
         end
         attr_writer :privacy_settings
 
+        # IDs of missions related to this assistant.
+        sig { returns(T.nilable(T::Array[String])) }
+        attr_reader :related_mission_ids
+
+        sig { params(related_mission_ids: T::Array[String]).void }
+        attr_writer :related_mission_ids
+
+        # Tags associated with the assistant. Tags can also be managed with the assistant
+        # tag endpoints.
+        sig { returns(T.nilable(T::Array[String])) }
+        attr_reader :tags
+
+        sig { params(tags: T::Array[String]).void }
+        attr_writer :tags
+
         sig { returns(T.nilable(Telnyx::AI::TelephonySettings)) }
         attr_reader :telephony_settings
 
@@ -164,8 +268,9 @@ module Telnyx
         end
         attr_writer :telephony_settings
 
-        # The tools that the assistant can use. These may be templated with
-        # [dynamic variables](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
+        # Deprecated for new integrations. Inline tool definitions available to the
+        # assistant. Prefer `tool_ids` to attach shared tools created with the AI Tools
+        # endpoints.
         sig do
           returns(T.nilable(T::Array[Telnyx::AI::AssistantTool::Variants]))
         end
@@ -200,6 +305,28 @@ module Telnyx
         end
         attr_writer :transcription
 
+        # Timestamp when this assistant version was created.
+        sig { returns(T.nilable(Time)) }
+        attr_reader :version_created_at
+
+        sig { params(version_created_at: Time).void }
+        attr_writer :version_created_at
+
+        # Identifier for the assistant version returned by version-aware assistant
+        # endpoints.
+        sig { returns(T.nilable(String)) }
+        attr_reader :version_id
+
+        sig { params(version_id: String).void }
+        attr_writer :version_id
+
+        # Human-readable name for the assistant version.
+        sig { returns(T.nilable(String)) }
+        attr_reader :version_name
+
+        sig { params(version_name: String).void }
+        attr_writer :version_name
+
         sig { returns(T.nilable(Telnyx::AI::VoiceSettings)) }
         attr_reader :voice_settings
 
@@ -222,19 +349,29 @@ module Telnyx
             name: String,
             description: String,
             dynamic_variables: T::Hash[Symbol, T.anything],
+            dynamic_variables_webhook_timeout_ms: Integer,
             dynamic_variables_webhook_url: String,
             enabled_features: T::Array[Telnyx::AI::EnabledFeatures::OrSymbol],
-            external_llm: Telnyx::AI::ExternalLlm::OrHash,
-            fallback_config: Telnyx::AI::FallbackConfig::OrHash,
+            external_llm: Telnyx::AI::InferenceEmbedding::ExternalLlm::OrHash,
+            fallback_config:
+              Telnyx::AI::InferenceEmbedding::FallbackConfig::OrHash,
             greeting: String,
             import_metadata: Telnyx::AI::ImportMetadata::OrHash,
             insight_settings: Telnyx::AI::InsightSettings::OrHash,
+            integrations:
+              T::Array[Telnyx::AI::InferenceEmbedding::Integration::OrHash],
+            interruption_settings:
+              Telnyx::AI::InferenceEmbedding::InterruptionSettings::OrHash,
             llm_api_key_ref: String,
+            mcp_servers:
+              T::Array[Telnyx::AI::InferenceEmbedding::McpServer::OrHash],
             messaging_settings: Telnyx::AI::MessagingSettings::OrHash,
             observability_settings: Telnyx::AI::Observability::OrHash,
             post_conversation_settings:
-              Telnyx::AI::PostConversationSettings::OrHash,
+              Telnyx::AI::InferenceEmbedding::PostConversationSettings::OrHash,
             privacy_settings: Telnyx::AI::PrivacySettings::OrHash,
+            related_mission_ids: T::Array[String],
+            tags: T::Array[String],
             telephony_settings: Telnyx::AI::TelephonySettings::OrHash,
             tools:
               T::Array[
@@ -252,6 +389,9 @@ module Telnyx
                 )
               ],
             transcription: Telnyx::AI::TranscriptionSettings::OrHash,
+            version_created_at: Time,
+            version_id: String,
+            version_name: String,
             voice_settings: Telnyx::AI::VoiceSettings::OrHash,
             widget_settings: Telnyx::AI::WidgetSettings::OrHash
           ).returns(T.attached_class)
@@ -262,18 +402,29 @@ module Telnyx
           # System instructions for the assistant. These may be templated with
           # [dynamic variables](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
           instructions:,
-          # ID of the model to use. You can use the
+          # ID of the model to use when `external_llm` is not set. You can use the
           # [Get models API](https://developers.telnyx.com/api-reference/chat/get-available-models)
-          # to see all of your available models,
+          # to see available models. If `external_llm` is provided, the assistant uses
+          # `external_llm` instead of this field. If neither `model` nor `external_llm` is
+          # provided, Telnyx applies the default model.
           model:,
           name:,
           description: nil,
           # Map of dynamic variables and their values
           dynamic_variables: nil,
-          # If the dynamic_variables_webhook_url is set for the assistant, we will send a
-          # request at the start of the conversation. See our
-          # [guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
-          # for more information.
+          # Timeout in milliseconds for the dynamic variables webhook. Must be between 1 and
+          # 10000 ms. If the webhook does not respond within this timeout, the call proceeds
+          # with default values. See the
+          # [dynamic variables guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables).
+          dynamic_variables_webhook_timeout_ms: nil,
+          # If `dynamic_variables_webhook_url` is set, Telnyx sends a POST request to this
+          # URL at the start of the conversation to resolve dynamic variables. **Gotcha:**
+          # the webhook response must wrap variables under a top-level `dynamic_variables`
+          # object, e.g. `{"dynamic_variables": {"customer_name": "Jane"}}`. Returning a
+          # flat object will be ignored and variables will fall back to their defaults. See
+          # the
+          # [dynamic variables guide](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
+          # for the full request/response format and timeout behavior.
           dynamic_variables_webhook_url: nil,
           enabled_features: nil,
           external_llm: nil,
@@ -287,12 +438,28 @@ module Telnyx
           greeting: nil,
           import_metadata: nil,
           insight_settings: nil,
-          # This is only needed when using third-party inference providers. The `identifier`
-          # for an integration secret
+          # Connected integrations attached to the assistant. The catalog of available
+          # integrations is at `/ai/integrations`; the user's connected integrations are at
+          # `/ai/integrations/connections`. Each item references a catalog integration by
+          # `integration_id`.
+          integrations: nil,
+          # Settings for interruptions and how the assistant decides the user has finished
+          # speaking. These timings are most relevant when using non turn-taking
+          # transcription models. For turn-taking models like `deepgram/flux`, end-of-turn
+          # behavior is controlled by the transcription end-of-turn settings under
+          # `transcription.settings` (`eot_threshold`, `eot_timeout_ms`,
+          # `eager_eot_threshold`).
+          interruption_settings: nil,
+          # This is only needed when using third-party inference providers selected by
+          # `model`. The `identifier` for an integration secret
           # [/v2/integration_secrets](https://developers.telnyx.com/api-reference/integration-secrets/create-a-secret)
-          # that refers to your LLM provider's API key. Warning: Free plans are unlikely to
-          # work with this integration.
+          # that refers to your LLM provider's API key. For bring-your-own endpoint
+          # authentication, use `external_llm.llm_api_key_ref` instead. Warning: Free plans
+          # are unlikely to work with this integration.
           llm_api_key_ref: nil,
+          # MCP servers attached to the assistant. Create MCP servers with
+          # `/ai/mcp_servers`, then reference them by `id` here.
+          mcp_servers: nil,
           messaging_settings: nil,
           observability_settings: nil,
           # Configuration for post-conversation processing. When enabled, the assistant
@@ -303,11 +470,24 @@ module Telnyx
           # post-conversation. Beta feature.
           post_conversation_settings: nil,
           privacy_settings: nil,
+          # IDs of missions related to this assistant.
+          related_mission_ids: nil,
+          # Tags associated with the assistant. Tags can also be managed with the assistant
+          # tag endpoints.
+          tags: nil,
           telephony_settings: nil,
-          # The tools that the assistant can use. These may be templated with
-          # [dynamic variables](https://developers.telnyx.com/docs/inference/ai-assistants/dynamic-variables)
+          # Deprecated for new integrations. Inline tool definitions available to the
+          # assistant. Prefer `tool_ids` to attach shared tools created with the AI Tools
+          # endpoints.
           tools: nil,
           transcription: nil,
+          # Timestamp when this assistant version was created.
+          version_created_at: nil,
+          # Identifier for the assistant version returned by version-aware assistant
+          # endpoints.
+          version_id: nil,
+          # Human-readable name for the assistant version.
+          version_name: nil,
           voice_settings: nil,
           # Configuration settings for the assistant's web widget.
           widget_settings: nil
@@ -324,22 +504,34 @@ module Telnyx
               name: String,
               description: String,
               dynamic_variables: T::Hash[Symbol, T.anything],
+              dynamic_variables_webhook_timeout_ms: Integer,
               dynamic_variables_webhook_url: String,
               enabled_features:
                 T::Array[Telnyx::AI::EnabledFeatures::TaggedSymbol],
-              external_llm: Telnyx::AI::ExternalLlm,
-              fallback_config: Telnyx::AI::FallbackConfig,
+              external_llm: Telnyx::AI::InferenceEmbedding::ExternalLlm,
+              fallback_config: Telnyx::AI::InferenceEmbedding::FallbackConfig,
               greeting: String,
               import_metadata: Telnyx::AI::ImportMetadata,
               insight_settings: Telnyx::AI::InsightSettings,
+              integrations:
+                T::Array[Telnyx::AI::InferenceEmbedding::Integration],
+              interruption_settings:
+                Telnyx::AI::InferenceEmbedding::InterruptionSettings,
               llm_api_key_ref: String,
+              mcp_servers: T::Array[Telnyx::AI::InferenceEmbedding::McpServer],
               messaging_settings: Telnyx::AI::MessagingSettings,
               observability_settings: Telnyx::AI::Observability,
-              post_conversation_settings: Telnyx::AI::PostConversationSettings,
+              post_conversation_settings:
+                Telnyx::AI::InferenceEmbedding::PostConversationSettings,
               privacy_settings: Telnyx::AI::PrivacySettings,
+              related_mission_ids: T::Array[String],
+              tags: T::Array[String],
               telephony_settings: Telnyx::AI::TelephonySettings,
               tools: T::Array[Telnyx::AI::AssistantTool::Variants],
               transcription: Telnyx::AI::TranscriptionSettings,
+              version_created_at: Time,
+              version_id: String,
+              version_name: String,
               voice_settings: Telnyx::AI::VoiceSettings,
               widget_settings: Telnyx::AI::WidgetSettings
             }
@@ -347,6 +539,750 @@ module Telnyx
         end
         def to_hash
         end
+
+        class ExternalLlm < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::ExternalLlm,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          # Base URL for the external LLM endpoint.
+          sig { returns(String) }
+          attr_accessor :base_url
+
+          # Model identifier to use with the external LLM endpoint.
+          sig { returns(String) }
+          attr_accessor :model
+
+          # Authentication method used when connecting to the external LLM endpoint.
+          sig do
+            returns(
+              T.nilable(
+                Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::TaggedSymbol
+              )
+            )
+          end
+          attr_reader :authentication_method
+
+          sig do
+            params(
+              authentication_method:
+                Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::OrSymbol
+            ).void
+          end
+          attr_writer :authentication_method
+
+          # Integration secret identifier for the client certificate used with certificate
+          # authentication.
+          sig { returns(T.nilable(String)) }
+          attr_reader :certificate_ref
+
+          sig { params(certificate_ref: String).void }
+          attr_writer :certificate_ref
+
+          # When `true`, Telnyx forwards the assistant's dynamic variables to the external
+          # LLM endpoint as a top-level `extra_metadata` object on the chat completion
+          # request body. Defaults to `false`. Example payload sent to the external
+          # endpoint:
+          # `{"extra_metadata": {"customer_name": "Jane", "account_id": "acct_789", "telnyx_agent_target": "+13125550100", "telnyx_end_user_target": "+13125550123"}}`.
+          # Distinct from OpenAI's native `metadata` field, which has its own size and type
+          # limits.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :forward_metadata
+
+          sig { params(forward_metadata: T::Boolean).void }
+          attr_writer :forward_metadata
+
+          # Integration secret identifier for the external LLM API key.
+          sig { returns(T.nilable(String)) }
+          attr_reader :llm_api_key_ref
+
+          sig { params(llm_api_key_ref: String).void }
+          attr_writer :llm_api_key_ref
+
+          # URL used to retrieve an access token when certificate authentication is enabled.
+          sig { returns(T.nilable(String)) }
+          attr_reader :token_retrieval_url
+
+          sig { params(token_retrieval_url: String).void }
+          attr_writer :token_retrieval_url
+
+          sig do
+            params(
+              base_url: String,
+              model: String,
+              authentication_method:
+                Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::OrSymbol,
+              certificate_ref: String,
+              forward_metadata: T::Boolean,
+              llm_api_key_ref: String,
+              token_retrieval_url: String
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Base URL for the external LLM endpoint.
+            base_url:,
+            # Model identifier to use with the external LLM endpoint.
+            model:,
+            # Authentication method used when connecting to the external LLM endpoint.
+            authentication_method: nil,
+            # Integration secret identifier for the client certificate used with certificate
+            # authentication.
+            certificate_ref: nil,
+            # When `true`, Telnyx forwards the assistant's dynamic variables to the external
+            # LLM endpoint as a top-level `extra_metadata` object on the chat completion
+            # request body. Defaults to `false`. Example payload sent to the external
+            # endpoint:
+            # `{"extra_metadata": {"customer_name": "Jane", "account_id": "acct_789", "telnyx_agent_target": "+13125550100", "telnyx_end_user_target": "+13125550123"}}`.
+            # Distinct from OpenAI's native `metadata` field, which has its own size and type
+            # limits.
+            forward_metadata: nil,
+            # Integration secret identifier for the external LLM API key.
+            llm_api_key_ref: nil,
+            # URL used to retrieve an access token when certificate authentication is enabled.
+            token_retrieval_url: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              {
+                base_url: String,
+                model: String,
+                authentication_method:
+                  Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::TaggedSymbol,
+                certificate_ref: String,
+                forward_metadata: T::Boolean,
+                llm_api_key_ref: String,
+                token_retrieval_url: String
+              }
+            )
+          end
+          def to_hash
+          end
+
+          # Authentication method used when connecting to the external LLM endpoint.
+          module AuthenticationMethod
+            extend Telnyx::Internal::Type::Enum
+
+            TaggedSymbol =
+              T.type_alias do
+                T.all(
+                  Symbol,
+                  Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod
+                )
+              end
+            OrSymbol = T.type_alias { T.any(Symbol, String) }
+
+            TOKEN =
+              T.let(
+                :token,
+                Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::TaggedSymbol
+              )
+            CERTIFICATE =
+              T.let(
+                :certificate,
+                Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::TaggedSymbol
+              )
+
+            sig do
+              override.returns(
+                T::Array[
+                  Telnyx::AI::InferenceEmbedding::ExternalLlm::AuthenticationMethod::TaggedSymbol
+                ]
+              )
+            end
+            def self.values
+            end
+          end
+        end
+
+        class FallbackConfig < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::FallbackConfig,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          sig do
+            returns(
+              T.nilable(
+                Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm
+              )
+            )
+          end
+          attr_reader :external_llm
+
+          sig do
+            params(
+              external_llm:
+                Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::OrHash
+            ).void
+          end
+          attr_writer :external_llm
+
+          # Integration secret identifier for the fallback model API key.
+          sig { returns(T.nilable(String)) }
+          attr_reader :llm_api_key_ref
+
+          sig { params(llm_api_key_ref: String).void }
+          attr_writer :llm_api_key_ref
+
+          # Fallback Telnyx-hosted model to use when the primary LLM provider is
+          # unavailable.
+          sig { returns(T.nilable(String)) }
+          attr_reader :model
+
+          sig { params(model: String).void }
+          attr_writer :model
+
+          sig do
+            params(
+              external_llm:
+                Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::OrHash,
+              llm_api_key_ref: String,
+              model: String
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            external_llm: nil,
+            # Integration secret identifier for the fallback model API key.
+            llm_api_key_ref: nil,
+            # Fallback Telnyx-hosted model to use when the primary LLM provider is
+            # unavailable.
+            model: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              {
+                external_llm:
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm,
+                llm_api_key_ref: String,
+                model: String
+              }
+            )
+          end
+          def to_hash
+          end
+
+          class ExternalLlm < Telnyx::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm,
+                  Telnyx::Internal::AnyHash
+                )
+              end
+
+            # Base URL for the external LLM endpoint.
+            sig { returns(String) }
+            attr_accessor :base_url
+
+            # Model identifier to use with the external LLM endpoint.
+            sig { returns(String) }
+            attr_accessor :model
+
+            # Authentication method used when connecting to the external LLM endpoint.
+            sig do
+              returns(
+                T.nilable(
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::TaggedSymbol
+                )
+              )
+            end
+            attr_reader :authentication_method
+
+            sig do
+              params(
+                authentication_method:
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::OrSymbol
+              ).void
+            end
+            attr_writer :authentication_method
+
+            # Integration secret identifier for the client certificate used with certificate
+            # authentication.
+            sig { returns(T.nilable(String)) }
+            attr_reader :certificate_ref
+
+            sig { params(certificate_ref: String).void }
+            attr_writer :certificate_ref
+
+            # When `true`, Telnyx forwards the assistant's dynamic variables to the external
+            # LLM endpoint as a top-level `extra_metadata` object on the chat completion
+            # request body. Defaults to `false`. Example payload sent to the external
+            # endpoint:
+            # `{"extra_metadata": {"customer_name": "Jane", "account_id": "acct_789", "telnyx_agent_target": "+13125550100", "telnyx_end_user_target": "+13125550123"}}`.
+            # Distinct from OpenAI's native `metadata` field, which has its own size and type
+            # limits.
+            sig { returns(T.nilable(T::Boolean)) }
+            attr_reader :forward_metadata
+
+            sig { params(forward_metadata: T::Boolean).void }
+            attr_writer :forward_metadata
+
+            # Integration secret identifier for the external LLM API key.
+            sig { returns(T.nilable(String)) }
+            attr_reader :llm_api_key_ref
+
+            sig { params(llm_api_key_ref: String).void }
+            attr_writer :llm_api_key_ref
+
+            # URL used to retrieve an access token when certificate authentication is enabled.
+            sig { returns(T.nilable(String)) }
+            attr_reader :token_retrieval_url
+
+            sig { params(token_retrieval_url: String).void }
+            attr_writer :token_retrieval_url
+
+            sig do
+              params(
+                base_url: String,
+                model: String,
+                authentication_method:
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::OrSymbol,
+                certificate_ref: String,
+                forward_metadata: T::Boolean,
+                llm_api_key_ref: String,
+                token_retrieval_url: String
+              ).returns(T.attached_class)
+            end
+            def self.new(
+              # Base URL for the external LLM endpoint.
+              base_url:,
+              # Model identifier to use with the external LLM endpoint.
+              model:,
+              # Authentication method used when connecting to the external LLM endpoint.
+              authentication_method: nil,
+              # Integration secret identifier for the client certificate used with certificate
+              # authentication.
+              certificate_ref: nil,
+              # When `true`, Telnyx forwards the assistant's dynamic variables to the external
+              # LLM endpoint as a top-level `extra_metadata` object on the chat completion
+              # request body. Defaults to `false`. Example payload sent to the external
+              # endpoint:
+              # `{"extra_metadata": {"customer_name": "Jane", "account_id": "acct_789", "telnyx_agent_target": "+13125550100", "telnyx_end_user_target": "+13125550123"}}`.
+              # Distinct from OpenAI's native `metadata` field, which has its own size and type
+              # limits.
+              forward_metadata: nil,
+              # Integration secret identifier for the external LLM API key.
+              llm_api_key_ref: nil,
+              # URL used to retrieve an access token when certificate authentication is enabled.
+              token_retrieval_url: nil
+            )
+            end
+
+            sig do
+              override.returns(
+                {
+                  base_url: String,
+                  model: String,
+                  authentication_method:
+                    Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::TaggedSymbol,
+                  certificate_ref: String,
+                  forward_metadata: T::Boolean,
+                  llm_api_key_ref: String,
+                  token_retrieval_url: String
+                }
+              )
+            end
+            def to_hash
+            end
+
+            # Authentication method used when connecting to the external LLM endpoint.
+            module AuthenticationMethod
+              extend Telnyx::Internal::Type::Enum
+
+              TaggedSymbol =
+                T.type_alias do
+                  T.all(
+                    Symbol,
+                    Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod
+                  )
+                end
+              OrSymbol = T.type_alias { T.any(Symbol, String) }
+
+              TOKEN =
+                T.let(
+                  :token,
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::TaggedSymbol
+                )
+              CERTIFICATE =
+                T.let(
+                  :certificate,
+                  Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::TaggedSymbol
+                )
+
+              sig do
+                override.returns(
+                  T::Array[
+                    Telnyx::AI::InferenceEmbedding::FallbackConfig::ExternalLlm::AuthenticationMethod::TaggedSymbol
+                  ]
+                )
+              end
+              def self.values
+              end
+            end
+          end
+        end
+
+        class Integration < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::Integration,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          # Catalog integration ID to attach. This is the `id` from the integrations catalog
+          # at `/ai/integrations` (the same value also appears as `integration_id` on
+          # entries returned by `/ai/integrations/connections`). It is **not** the
+          # connection-level `id` from `/ai/integrations/connections`.
+          sig { returns(String) }
+          attr_accessor :integration_id
+
+          # Optional per-assistant allowlist of integration tool names. When omitted or
+          # empty, all tools allowed by the connected integration are available to the
+          # assistant.
+          sig { returns(T.nilable(T::Array[String])) }
+          attr_reader :allowed_list
+
+          sig { params(allowed_list: T::Array[String]).void }
+          attr_writer :allowed_list
+
+          # Reference to a connected integration attached to an assistant. Discover
+          # available integrations with `/ai/integrations` and connected integrations with
+          # `/ai/integrations/connections`.
+          sig do
+            params(
+              integration_id: String,
+              allowed_list: T::Array[String]
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Catalog integration ID to attach. This is the `id` from the integrations catalog
+            # at `/ai/integrations` (the same value also appears as `integration_id` on
+            # entries returned by `/ai/integrations/connections`). It is **not** the
+            # connection-level `id` from `/ai/integrations/connections`.
+            integration_id:,
+            # Optional per-assistant allowlist of integration tool names. When omitted or
+            # empty, all tools allowed by the connected integration are available to the
+            # assistant.
+            allowed_list: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              { integration_id: String, allowed_list: T::Array[String] }
+            )
+          end
+          def to_hash
+          end
+        end
+
+        class InterruptionSettings < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::InterruptionSettings,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          # Whether users can interrupt the assistant while it is speaking.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :enable
+
+          sig { params(enable: T::Boolean).void }
+          attr_writer :enable
+
+          # Controls when the assistant starts speaking after the user stops. These
+          # thresholds primarily apply to non turn-taking transcription models. For
+          # turn-taking models like `deepgram/flux`, end-of-turn detection is driven by the
+          # transcription end-of-turn settings under `transcription.settings` instead.
+          sig do
+            returns(
+              T.nilable(
+                Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan
+              )
+            )
+          end
+          attr_reader :start_speaking_plan
+
+          sig do
+            params(
+              start_speaking_plan:
+                Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::OrHash
+            ).void
+          end
+          attr_writer :start_speaking_plan
+
+          # Settings for interruptions and how the assistant decides the user has finished
+          # speaking. These timings are most relevant when using non turn-taking
+          # transcription models. For turn-taking models like `deepgram/flux`, end-of-turn
+          # behavior is controlled by the transcription end-of-turn settings under
+          # `transcription.settings` (`eot_threshold`, `eot_timeout_ms`,
+          # `eager_eot_threshold`).
+          sig do
+            params(
+              enable: T::Boolean,
+              start_speaking_plan:
+                Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::OrHash
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Whether users can interrupt the assistant while it is speaking.
+            enable: nil,
+            # Controls when the assistant starts speaking after the user stops. These
+            # thresholds primarily apply to non turn-taking transcription models. For
+            # turn-taking models like `deepgram/flux`, end-of-turn detection is driven by the
+            # transcription end-of-turn settings under `transcription.settings` instead.
+            start_speaking_plan: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              {
+                enable: T::Boolean,
+                start_speaking_plan:
+                  Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan
+              }
+            )
+          end
+          def to_hash
+          end
+
+          class StartSpeakingPlan < Telnyx::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan,
+                  Telnyx::Internal::AnyHash
+                )
+              end
+
+            # Endpointing thresholds used to decide when the user has finished speaking.
+            # Applies to non turn-taking transcription models. For `deepgram/flux`, use
+            # `transcription.settings.eot_threshold` / `eot_timeout_ms` /
+            # `eager_eot_threshold`.
+            sig do
+              returns(
+                T.nilable(
+                  Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::TranscriptionEndpointingPlan
+                )
+              )
+            end
+            attr_reader :transcription_endpointing_plan
+
+            sig do
+              params(
+                transcription_endpointing_plan:
+                  Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::TranscriptionEndpointingPlan::OrHash
+              ).void
+            end
+            attr_writer :transcription_endpointing_plan
+
+            # Minimum seconds to wait before the assistant starts speaking.
+            sig { returns(T.nilable(Float)) }
+            attr_reader :wait_seconds
+
+            sig { params(wait_seconds: Float).void }
+            attr_writer :wait_seconds
+
+            # Controls when the assistant starts speaking after the user stops. These
+            # thresholds primarily apply to non turn-taking transcription models. For
+            # turn-taking models like `deepgram/flux`, end-of-turn detection is driven by the
+            # transcription end-of-turn settings under `transcription.settings` instead.
+            sig do
+              params(
+                transcription_endpointing_plan:
+                  Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::TranscriptionEndpointingPlan::OrHash,
+                wait_seconds: Float
+              ).returns(T.attached_class)
+            end
+            def self.new(
+              # Endpointing thresholds used to decide when the user has finished speaking.
+              # Applies to non turn-taking transcription models. For `deepgram/flux`, use
+              # `transcription.settings.eot_threshold` / `eot_timeout_ms` /
+              # `eager_eot_threshold`.
+              transcription_endpointing_plan: nil,
+              # Minimum seconds to wait before the assistant starts speaking.
+              wait_seconds: nil
+            )
+            end
+
+            sig do
+              override.returns(
+                {
+                  transcription_endpointing_plan:
+                    Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::TranscriptionEndpointingPlan,
+                  wait_seconds: Float
+                }
+              )
+            end
+            def to_hash
+            end
+
+            class TranscriptionEndpointingPlan < Telnyx::Internal::Type::BaseModel
+              OrHash =
+                T.type_alias do
+                  T.any(
+                    Telnyx::AI::InferenceEmbedding::InterruptionSettings::StartSpeakingPlan::TranscriptionEndpointingPlan,
+                    Telnyx::Internal::AnyHash
+                  )
+                end
+
+              # Seconds to wait after the transcript ends without punctuation.
+              sig { returns(T.nilable(Float)) }
+              attr_reader :on_no_punctuation_seconds
+
+              sig { params(on_no_punctuation_seconds: Float).void }
+              attr_writer :on_no_punctuation_seconds
+
+              # Seconds to wait after the transcript ends with a number.
+              sig { returns(T.nilable(Float)) }
+              attr_reader :on_number_seconds
+
+              sig { params(on_number_seconds: Float).void }
+              attr_writer :on_number_seconds
+
+              # Seconds to wait after the transcript ends with punctuation.
+              sig { returns(T.nilable(Float)) }
+              attr_reader :on_punctuation_seconds
+
+              sig { params(on_punctuation_seconds: Float).void }
+              attr_writer :on_punctuation_seconds
+
+              # Endpointing thresholds used to decide when the user has finished speaking.
+              # Applies to non turn-taking transcription models. For `deepgram/flux`, use
+              # `transcription.settings.eot_threshold` / `eot_timeout_ms` /
+              # `eager_eot_threshold`.
+              sig do
+                params(
+                  on_no_punctuation_seconds: Float,
+                  on_number_seconds: Float,
+                  on_punctuation_seconds: Float
+                ).returns(T.attached_class)
+              end
+              def self.new(
+                # Seconds to wait after the transcript ends without punctuation.
+                on_no_punctuation_seconds: nil,
+                # Seconds to wait after the transcript ends with a number.
+                on_number_seconds: nil,
+                # Seconds to wait after the transcript ends with punctuation.
+                on_punctuation_seconds: nil
+              )
+              end
+
+              sig do
+                override.returns(
+                  {
+                    on_no_punctuation_seconds: Float,
+                    on_number_seconds: Float,
+                    on_punctuation_seconds: Float
+                  }
+                )
+              end
+              def to_hash
+              end
+            end
+          end
+        end
+
+        class McpServer < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::McpServer,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          # ID of the MCP server to attach. This must be the `id` of an MCP server returned
+          # by the `/ai/mcp_servers` endpoints.
+          sig { returns(String) }
+          attr_accessor :id
+
+          # Optional per-assistant allowlist of MCP tool names. When omitted, the assistant
+          # uses the MCP server's configured `allowed_tools`.
+          sig { returns(T.nilable(T::Array[String])) }
+          attr_reader :allowed_tools
+
+          sig { params(allowed_tools: T::Array[String]).void }
+          attr_writer :allowed_tools
+
+          # Reference to an MCP server attached to an assistant. Create and manage MCP
+          # servers with the `/ai/mcp_servers` endpoints, then attach them to assistants by
+          # ID.
+          sig do
+            params(id: String, allowed_tools: T::Array[String]).returns(
+              T.attached_class
+            )
+          end
+          def self.new(
+            # ID of the MCP server to attach. This must be the `id` of an MCP server returned
+            # by the `/ai/mcp_servers` endpoints.
+            id:,
+            # Optional per-assistant allowlist of MCP tool names. When omitted, the assistant
+            # uses the MCP server's configured `allowed_tools`.
+            allowed_tools: nil
+          )
+          end
+
+          sig do
+            override.returns({ id: String, allowed_tools: T::Array[String] })
+          end
+          def to_hash
+          end
+        end
+
+        class PostConversationSettings < Telnyx::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                Telnyx::AI::InferenceEmbedding::PostConversationSettings,
+                Telnyx::Internal::AnyHash
+              )
+            end
+
+          # Whether post-conversation processing is enabled. When true, the assistant will
+          # be invoked after the conversation ends to perform any final tool calls. Defaults
+          # to false.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :enabled
+
+          sig { params(enabled: T::Boolean).void }
+          attr_writer :enabled
+
+          # Configuration for post-conversation processing. When enabled, the assistant
+          # receives one additional LLM turn after the conversation ends, allowing it to
+          # execute tool calls such as logging to a CRM or sending a summary. The assistant
+          # can execute multiple parallel or sequential tools during this phase.
+          # Telephony-control tools (e.g. hangup, transfer) are unavailable
+          # post-conversation. Beta feature.
+          sig { params(enabled: T::Boolean).returns(T.attached_class) }
+          def self.new(
+            # Whether post-conversation processing is enabled. When true, the assistant will
+            # be invoked after the conversation ends to perform any final tool calls. Defaults
+            # to false.
+            enabled: nil
+          )
+          end
+
+          sig { override.returns({ enabled: T::Boolean }) }
+          def to_hash
+          end
+        end
       end
     end
   end