TypeSafe Jev Decision Model

Submit content together with a set of questions to receive structured probabilities, choices, or scores. Use cases include intent classification, ticket routing, content moderation, and quality scoring. All questions in a request share the same state.

Endpoint

Method	Endpoint
POST	https://neolink.com/api/v1/decisions

Compatibility path: https://neolink.com/api/v1/systemone, with the same request and response format. Use /v1/decisions for new integrations.

Authentication

Method	Description
Bearer Token	Include Authorization: Bearer <API Key> in the request header. Use an API Key created on this platform.

Request Headers

Parameter	Type	Required	Description
Content-Type	string	Yes	Set to application/json
Authorization	string	Yes	Bearer <API Key> for account authentication

Model Capabilities

• Supports noul (binary judgment), choice (selection), and score (rating). Different question types can be combined in one request.
• Returns a complete JSON response. Streaming is not supported; do not set stream: true.
• Image input (assets) and box localization questions are not currently supported by this endpoint.

Model ID: Examples use jev-latest. Available models and access permissions depend on your platform configuration.

Request Body

Parameter	Type	Required	Default	Description
model	string	Yes	-	Model ID, such as jev-latest
state	string/object/array	Yes	-	Text, conversation history, data records, or application state to evaluate, shared by all questions
questions	object	Yes	-	Non-empty question map. Caller-defined keys identify entries in answers and do not participate in model reasoning
questions.<key>.type	string	Yes	-	Question type: noul, choice, or score
questions.<key>.instructions	string/object/array	Yes	-	Question description. Objects or arrays can separate the question from reference data; refer to fields in text using backticks
questions.<key>.criteria	object/array	Conditional	-	Evaluation criteria. See the formats and limits below
stream	boolean	No	false	Omit or set to false

criteria Parameters

Question Type	Type	Required	Description
noul	object	No	Use true and false keys to describe the two outcomes. Descriptions accept strings, objects, or arrays. Example: {"true": "Explicitly time-sensitive", "false": "No urgency expressed"}
choice	object	Yes	1–255 options. Keys are the possible returned choices; descriptions accept strings, objects, arrays, or null. Example: {"billing": "Payments and refunds", "technical": null}
score	array[string]	Yes	2–10 rating levels ordered from low to high, indexed from 0. Example: ["Calm", "Frustrated", "Very angry"]

Request Example

This example evaluates urgency, selects a department, and rates frustration for the same customer message. You can also submit just one of these questions.

cURL

curl --location --request POST 'https://neolink.com/api/v1/decisions' \
     --header 'Authorization: Bearer <API Key>' \
     --header 'Content-Type: application/json' \
     --data-raw '{
    "model": "jev-latest",
    "state": "Help! My payouts have been failing for 3 days.",
    "questions": {
        "is_urgent": {
            "type": "noul",
            "instructions": "Does this convey urgency?",
            "criteria": {
                "true": "Explicitly time-sensitive",
                "false": "No urgency expressed"
            }
        },
        "department": {
            "type": "choice",
            "instructions": "Which team should handle this?",
            "criteria": {
                "billing": "Payments, invoicing, refunds",
                "technical": "Bugs, outages, integrations",
                "sales": "Pricing, upgrades, new accounts"
            }
        },
        "frustration": {
            "type": "score",
            "instructions": "How frustrated is the customer?",
            "criteria": [
                "Calm",
                "Frustrated",
                "Very angry"
            ]
        }
    }
}'

Python

import requests

url = "https://neolink.com/api/v1/decisions"
headers = {
    "Authorization": "Bearer <API Key>",
    "Content-Type": "application/json",
}
payload = {
    "model": "jev-latest",
    "state": "Help! My payouts have been failing for 3 days.",
    "questions": {
        "is_urgent": {
            "type": "noul",
            "instructions": "Does this convey urgency?",
            "criteria": {
                "true": "Explicitly time-sensitive",
                "false": "No urgency expressed"
            }
        },
        "department": {
            "type": "choice",
            "instructions": "Which team should handle this?",
            "criteria": {
                "billing": "Payments, invoicing, refunds",
                "technical": "Bugs, outages, integrations",
                "sales": "Pricing, upgrades, new accounts"
            }
        },
        "frustration": {
            "type": "score",
            "instructions": "How frustrated is the customer?",
            "criteria": [
                "Calm",
                "Frustrated",
                "Very angry"
            ]
        }
    }
}

response = requests.post(url, headers=headers, json=payload, timeout=60)
response.raise_for_status()
print(response.json())

Response Parameters

Parameter	Type	Description
model	string	Model ID in the response
answers	object	Answer map with keys matching the request's questions
answers.<key>.type	string	Answer type: noul, choice, or score
answers.<key>.noul	number	Likelihood of “yes” for noul, from 0 to 1; not a boolean
answers.<key>.choice	string	Selected option key for choice
answers.<key>.score	number	Rating for score, possibly fractional, from 0 to the number of levels minus 1
answers.<key>.legend	object	Mapping of level indices to descriptions for score
answers.<key>.probabilities	object	Probability of each option or level for choice or score
answers.<key>.confidence	number	Confidence for choice or score; higher values indicate greater certainty
usage.input_tokens	integer	Input token count for the request
usage.output_tokens	integer	Output token count for the request

Response Example

This illustrates the response structure. Actual results and token usage depend on the response returned by the API.

{
    "model": "jev-latest",
    "answers": {
        "is_urgent": {
            "type": "noul",
            "noul": 0.95
        },
        "department": {
            "type": "choice",
            "choice": "billing",
            "probabilities": {
                "billing": 0.88,
                "technical": 0.12,
                "sales": 0
            },
            "confidence": 0.81
        },
        "frustration": {
            "type": "score",
            "score": 1.05,
            "legend": {
                "0": "Calm",
                "1": "Frustrated",
                "2": "Very angry"
            },
            "probabilities": {
                "0": 0,
                "1": 0.95,
                "2": 0.05
            },
            "confidence": 0.92
        }
    },
    "usage": {
        "input_tokens": 450,
        "output_tokens": 72
    }
}

Usage Notes

• Choose a noul threshold for your application, for example flagging a message as urgent when noul >= 0.8. The caller determines the threshold.
• In the score example, 0, 1, and 2 correspond to the three frustration levels. 1.05 is a rating, not a percentage.
• Billing uses the returned token counts and the platform's configured rates. Refer to the prices displayed on the platform.
• For failed requests, see the Error Code Guide.