Update an assistant

  • Updated on Jan 7, 2026
  • Published on Jan 7, 2026
  • 3 minute(s) read
  • RK
 Prev Next
Patch
/assistants/{id}

Updates an existing assistant in both the external provider and the local database using a partial configuration payload.

Security
API Key: CookieAuth
Cookie parameter nameauth-token

Authentication is handled via an auth-token HTTP-only cookie set by the /auth/login endpoint.

  • In a browser, you typically call /auth/login once and the cookie will be stored and automatically sent on subsequent API requests.
  • In API tools (Postman, curl, etc.) you can manually capture the cookie from the Set-Cookie header or configure cookie handling.

This spec models the cookie using apiKey-style auth for documentation purposes.

Path parameters
id
stringRequired

Assistant identifier (external provider ID or local DB ID).

Body parameters
Expand All
object

Partial update payload for an existing assistant. All fields are optional.

Example{ "name": "Enrollment Assistant", "firstMessage": "Hi, I'm your enrollment assistant. How can I help today?", "firstMessageMode": "assistant-speaks-first", "model": { "provider": "openai", "model": "gpt-4o", "temperature": 0.6 }, "voice": { "provider": "openai", "voiceId": "alloy" }, "maxDurationSeconds": 1800 }
name
string

Assistant display name.

firstMessage
string | null
firstMessageMode
string | null
Valid values[ "assistant-speaks-first", "assistant-waits-for-user", "assistant-speaks-first-with-model-generated-message" ]
model
object (ModelConfig) Required

Configuration for the underlying language model (OpenAI / Anthropic).

For non-technical readers, you can think of this as picking which "brain" the agent uses and how creative it should be.

provider
string Required
Valid values[ "openai", "anthropic" ]
model
string Required

Model name from the chosen provider.

Examplegpt-4o
temperature
number

Controls creativity. Lower values are more deterministic, higher values more creative and varied.

Minimum0.0
Maximum2.0
Example0.7
maxTokens
integer

Maximum number of tokens the model can generate for a response.

Example1024
voice
object (VoiceConfig) Required

Voice configuration for spoken responses when used in calls.

provider
string

Voice provider identifier.

Valid values[ "11labs", "azure", "cartesia", "deepgram", "playht", "openai", "edmo" ]
voiceId
string

Voice identifier within the chosen provider.

Examplealloy
transcriber
object (TranscriberConfig)

Speech-to-text configuration used to transcribe calls.

provider
string
Valid values[ "deepgram", "assemblyai", "whisper" ]
model
string | null

Optional model identifier, depending on the provider.

voicemailMessage
string | null
endCallMessage
string | null
endCallPhrases
Array of string | null
string
metadata
object | null
maxDurationSeconds
integer | null
Responses
200

Assistant updated successfully.

Expand All
object

A voice/chat agent configuration that defines how the assistant behaves and which model, voice and tools it uses.

Example{ "id": "asst_123", "orgId": "org_abc", "createdAt": "2024-01-01T10:00:00Z", "updatedAt": "2024-01-02T12:00:00Z", "name": "Enrollment Assistant", "firstMessage": "Hi, I'm here to help with your enrollment questions.", "firstMessageMode": "assistant-speaks-first", "model": { "provider": "openai", "model": "gpt-4o", "temperature": 0.7 }, "voice": { "provider": "openai", "voiceId": "alloy" }, "maxDurationSeconds": 1800 }
id
string

Assistant identifier.

orgId
string

Owning organization identifier.

createdAt
string (date-time)
updatedAt
string (date-time)
name
string | null

Human-friendly assistant name shown in the dashboard.

firstMessage
string | null

First message sent by the assistant when a call/chat starts.

firstMessageMode
string | null
Valid values[ "assistant-speaks-first", "assistant-waits-for-user", "assistant-speaks-first-with-model-generated-message" ]
model
object (ModelConfig)

Configuration for the underlying language model (OpenAI / Anthropic).

For non-technical readers, you can think of this as picking which "brain" the agent uses and how creative it should be.

provider
string
Valid values[ "openai", "anthropic" ]
model
string

Model name from the chosen provider.

Examplegpt-4o
temperature
number

Controls creativity. Lower values are more deterministic, higher values more creative and varied.

Minimum0.0
Maximum2.0
Example0.7
maxTokens
integer

Maximum number of tokens the model can generate for a response.

Example1024
voice
object (VoiceConfig)

Voice configuration for spoken responses when used in calls.

provider
string

Voice provider identifier.

Valid values[ "11labs", "azure", "cartesia", "deepgram", "playht", "openai", "edmo" ]
voiceId
string

Voice identifier within the chosen provider.

Examplealloy
transcriber
object (TranscriberConfig)

Speech-to-text configuration used to transcribe calls.

provider
string
Valid values[ "deepgram", "assemblyai", "whisper" ]
model
string | null

Optional model identifier, depending on the provider.

voicemailMessage
string | null
endCallMessage
string | null
endCallPhrases
Array of string | null
string
metadata
object | null
maxDurationSeconds
integer | null
401

Not authenticated or lacking permission to update agents.

object

Standard error response envelope.

Example{ "error": "Failed to load agents. Please try again." }
error
string

Human-readable error message.

404

Assistant not found.

object

Standard error response envelope.

Example{ "error": "Failed to load agents. Please try again." }
error
string

Human-readable error message.

500

Error updating assistant or communicating with the external provider.

object

Standard error response envelope.

Example{ "error": "Failed to load agents. Please try again." }
error
string

Human-readable error message.