Reply Intent

The Reply Intent model is a fast and cheap model to classify a high volume of replies from leads.

On this page, we'll dive into the Reply Intent endpoints you can use to trigger the Reply Intent model programmatically.

The ReplyIntent type

A ReplyIntent is a type that contains information about the intent of a reply from a lead.

Required Properties

  • Name
    label
    Type
    enum['bounced', 'OOO', 'no-longer-works-here', 'not-interested', 'interested', 'wants-to-meet', 'meeting-booked', 'other']
    Description

    The label of the intent.

  • Name
    confidence
    Type
    number
    Description

    The confidence of the intent; between 0 and 1.


POST/v1/models/prediction

Initiate a Reply Intent run

This endpoint allows you to classify the intent of a reply from a lead.

Required attributes

  • Name
    model
    Type
    enum['reply-intent-light']
    Description

    The model to run. Must be reply-intent-light. The reply-intent model is coming soon.

  • Name
    input
    Type
    []string
    Description

    The replies to classify. Max. 32 replies per request.

Returns

  • Name
    model
    Type
    enum['reply-intent-light']
    Description

    The model that was used to classify the replies.

  • Name
    output
    Type
    []ReplyIntent
    Description

    The replies classified. See the ReplyIntent type for more information.

Request

POST
/v1/models/prediction
curl -X POST https://api.utopianlabs.ai/v1/models/prediction \
  -H "Authorization: Bearer {token}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "reply-intent-light",
    "input": [
      "Thanks for reaching out! I'd love to learn more.",
      "Please remove me from your list."
    ]
  }'

Response

{
  "model": "reply-intent-light",
  "output": [
    {
      "label": "interested",
      "confidence": 0.92
    },
    {
      "label": "not-interested",
      "confidence": 0.98
    }
  ]
}

Was this page helpful?