> ## Documentation Index > Fetch the complete documentation index at: https://docs.getoutbox.ai/llms.txt > Use this file to discover all available pages before exploring further. # Create Agent > Create a new agent (chatbot or voicebot) ## OpenAPI ````yaml POST /agent/ openapi: 3.1.0 info: title: Outbox AI API description: API for managing AI agents (chatbots and voicebots) on the Outbox platform version: 1.0.0 servers: - url: https://api.getoutbox.ai security: [] paths: /agent/: post: summary: Create Agent description: Create a new agent (chatbot or voicebot) requestBody: description: Agent creation details required: true content: application/json: schema: $ref: '#/components/schemas/AgentCreate' responses: '200': description: Agent created successfully content: application/json: schema: type: string description: The ID of the newly created agent '400': description: Bad Request - Agent limit reached or failed to create assistant '401': description: Unauthorized - Invalid or missing API key '403': description: Forbidden - Insufficient permissions (requires level 3) security: - CompanyApiKey: [] components: schemas: AgentCreate: oneOf: - $ref: '#/components/schemas/ChatAgentCreate' - $ref: '#/components/schemas/VoiceAgentCreate' ChatAgentCreate: title: Chat Agent type: object required: - name - type properties: name: type: string description: Name of the agent type: type: string enum: - chatbot description: Agent type discriminator assistant_id: type: string description: Internal assistant identifier (optional) description: type: string description: If present, used to generate the prompt prompt: type: string description: Used only if no description is provided first_message: type: string description: First message shown in supported chat surfaces timezone: type: string description: Timezone name (e.g. "America/New_York") webhook_url: type: string description: Webhook URL (if applicable) branding_colour: type: string description: '"" or hex #RGB / #RRGGBB' model: type: string default: gpt-4.1 description: Chat model name (defaults to gpt-4.1) VoiceAgentCreate: title: Voice Agent type: object required: - name - type properties: name: type: string description: Name of the agent type: type: string enum: - voicebot description: Agent type discriminator description: type: string description: If present, used to generate the prompt prompt: type: string description: Used only if no description is provided first_message: type: string description: First spoken message (blank means wait for caller) timezone: type: string description: Timezone name (e.g. "America/New_York") webhook_url: type: string description: Webhook URL to receive call completion payloads branding_colour: type: string description: '"" or hex #RGB / #RRGGBB' language: type: string description: Language code (e.g. "en") model: type: object description: LLM configuration for voice calls properties: model: type: string description: Model name max_tokens: type: integer description: Max tokens for responses temperature: type: number description: Sampling temperature voice: type: object description: Voice provider + voice id required: - provider - id properties: provider: type: string enum: - 11labs - openai description: Voice provider id: type: string description: Provider voice id model: type: string description: '11labs voice model (default: eleven_turbo_v2)' similarityBoost: type: number style: type: number stability: type: number optimizeStreamingLatency: type: integer useSpeakerBoost: type: boolean autoMode: type: boolean speed: type: number hipaa_enabled: description: Accepts boolean or boolean-ish strings oneOf: - type: boolean - type: string enum: - 'true' - 'false' zdr_enabled: type: boolean description: >- Enable Zero Data Retention where supported so no call data is stored for this agent recording_enabled: type: boolean description: Controls whether call audio recordings are stored transcript_enabled: type: boolean description: Controls whether call transcripts are stored voicemail_message: type: string description: What the agent says if voicemail is detected end_call_message: type: string description: Last line the agent says before hanging up recording_consent_message: type: string description: >- What the agent says to inform the caller the call may be recorded ("" clears) success_prompt: type: string description: Prompt used to evaluate whether the call achieved the agent's goal unqualified_prompt: type: string description: >- Prompt used to evaluate whether the caller is not a fit for your services scoring_prompt: type: string description: Prompt used to rate call success from 0 to 100 summary_prompt: type: string description: Prompt used to summarize the call keyterms: type: string description: >- Space-separated transcription hints (names, brands, locations) to improve speech-to-text accuracy ("" clears) background: type: string description: >- Background audio preset: "office", "lounge", or "off"; or a custom MP3 URL idle_messages: type: array items: type: string description: 'What to say when the caller is silent (default: none)' idle_message_max_spoken_count: type: integer minimum: 1 maximum: 10 description: Max number of idle messages per call idle_timeout_seconds: type: integer minimum: 0 maximum: 60 description: Seconds of silence before idle messages are eligible llm_request_delay_seconds: type: number minimum: 0 maximum: 10 max_call_duration_seconds: type: integer minimum: 10 maximum: 7200 interrupt_num_words: type: integer minimum: 0 maximum: 50 speech_end_of_turn: type: object properties: threshold: type: number minimum: 0 maximum: 1 timeout_ms: type: integer minimum: 0 maximum: 60000 speech_stop: type: object properties: num_words: type: integer minimum: 0 maximum: 50 voice_seconds: type: number minimum: 0 maximum: 10 backoff_seconds: type: number minimum: 0 maximum: 30 speech_start: type: object properties: wait_seconds: type: number minimum: 0 maximum: 10 securitySchemes: CompanyApiKey: type: apiKey in: header name: Authorization description: Company API Key ````