create_assistant
Create a new AI assistant. Required: name, voice_id, language_id, type, mode, timezone, initial_message, system_prompt, and either llm_model_id (pipeline) or multimodal_model_id (multimodal/dualplex). For multimodal/dualplex, knowledgebase_mode must be "function_call" when a knowledgebase is attached.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| mode | Yes | ||
| name | Yes | Display name of the assistant (max 255 chars) | |
| type | Yes | Assistant call direction | |
| tools | No | ||
| record | No | ||
| fillers | No | ||
| timezone | Yes | IANA timezone, e.g. "Europe/Berlin" | |
| tool_ids | No | ||
| voice_id | Yes | Voice ID from get_voices | |
| folder_id | No | Folder to place the assistant in (one folder per assistant). See list_folders. Send null to remove. | |
| label_ids | No | Label IDs to tag the assistant with (an assistant can carry multiple labels). See list_labels. | |
| variables | No | ||
| language_id | Yes | Primary language ID from get_languages | |
| webhook_url | No | ||
| llm_model_id | No | Required for mode=pipeline | |
| max_duration | No | ||
| ringing_time | No | ||
| speech_speed | No | ||
| ambient_sound | No | ||
| endpoint_type | No | ||
| filler_config | No | ||
| system_prompt | Yes | System prompt defining behavior | |
| initial_message | Yes | First message spoken at call start (max 200 chars) | |
| llm_temperature | No | ||
| phone_number_id | No | ||
| voice_stability | No | ||
| knowledgebase_id | No | ||
| post_call_schema | No | ||
| voice_similarity | No | ||
| is_webhook_active | No | ||
| wait_for_customer | No | ||
| knowledgebase_mode | No | ||
| voice_mail_message | No | ||
| allow_interruptions | No | ||
| min_interrupt_words | No | ||
| multimodal_model_id | No | Required for mode=multimodal/dualplex | |
| reengagement_prompt | No | ||
| tts_emotion_enabled | No | ||
| ambient_sound_volume | No | ||
| chat_llm_fallback_id | No | ||
| endpoint_sensitivity | No | ||
| max_silence_duration | No | ||
| post_call_evaluation | No | ||
| end_call_on_voicemail | No | ||
| interrupt_sensitivity | No | ||
| reengagement_interval | No | ||
| secondary_language_ids | No | Additional language IDs the assistant can speak | |
| synthesizer_provider_id | No | ||
| transcriber_provider_id | No | ||
| turn_detection_threshold | No | ||
| enable_noise_cancellation | No | ||
| conversation_ended_retrigger | No | ||
| include_recording_in_webhook | No | ||
| max_initial_silence_duration | No | ||
| conversation_ended_webhook_url | No | ||
| send_webhook_only_on_completed | No | ||
| conversation_inactivity_timeout | No |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| data | No | ||
| message | No |