Cold Email Evaluations

You can use the Cold Email Evaluations endpoint to score the performance of a specific cold email. You can do this to validate the cold-email-1 model or use this to get feedback on your (AI-)writting email.

On this page, we'll dive into the Cold Email Evaluation endpoint and how you can trigger it programmatically.

POST/v1/evaluations/cold-email

Run a Cold Email Evaluation

This endpoint allows you to evaluate an email.

Required attributes

Name
email
Type
string
Description
The email to evaluate, preferably in markdown or plain-text format.

Optional attributes

Name
context
Type
string
Description
Any context that is was used to write the email. This is mostly applicable when evaluating an AI written email, like when using cold-email-1.
Name
user
Type
EvaluationsInput
Description
More information about the user that's required for some of the evaluations. See the EvaluationsInput type for more information.
Name
lead
Type
EvaluationsInput
Description
More information about the lead that's required for some of the evaluations. See the EvaluationsInput type for more information.

Returns

Name
score
Type
number
Description
The overall score of the email
Name
evaluations
Type
EvaluationsArray
Description
An array containing the results of all of the evaluations that have been run. See the EvaluationsArray type for more information.

Request

POST

/v1/evaluations/cold-email

curl -X POST https://api.utopianlabs.ai/v1/evaluations/cold-email \
  -H "Authorization: Bearer {token}" \
  -H "Content-Type: application/json" \
  -d '{
    "email": "The email you want to evaluate",
    "lead": {
      "company_website": "https://example.com"
      "company_name": "Example Corp",
      "person_job_title": "CEO"
    },
    "user": {
      "company_website": "https://another-example.com"
      "company_name": "Another Example Corp",
      "person_name": "John Doe"
    },
    "context": "This lead asked me to contact him in 6 months. It has been 3 months now, is this the right time to contact them again?"
  }'

Response

{
  "score": 0.6817245,
  "evaluations": [
    {
      "key": "Name of the evaluation function",
      "score": 0.5,
      "feedback": "Any feedback that can help you improve the email."
    }
  ]
}