> ## Documentation Index
> Fetch the complete documentation index at: https://docs.paytring.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Beep Classification

> Turn any text or JSON state into typed decisions with calibrated confidence.

Beep is a non-autoregressive decision model. Instead of generating text, it answers **typed questions** about a *state* (a sentence, an email, a ticket, or any JSON document) in a single forward pass, and returns probabilities you can act on.

You define the decision at request time. No training, no prompt engineering, no output parsing.

## Base URL

```
https://ai.paytring.com/v1/classify/
```

<Note>
  **Authentication:** Send your Paytring credentials using **HTTP Basic authentication**. Use the same key and secret you use for other Paytring APIs. Contact your **Paytring account manager** to get access enabled for your account.
</Note>

```bash theme={null}
curl -u "$PAYTRING_KEY:$PAYTRING_SECRET" https://ai.paytring.com/v1/classify/ \
  -H 'content-type: application/json' \
  -d '{
    "state": "I was charged twice. Please refund the duplicate.",
    "questions": {
      "refund_requested": {
        "type": "noul",
        "instructions": "Does the customer request a refund?"
      }
    }
  }'
```

## What you can ask

| Type     | Criteria                                                  | You get back                                    |
| -------- | --------------------------------------------------------- | ----------------------------------------------- |
| `choice` | Label → description map (or list of labels), 1–20 options | Top label, per-label probabilities, confidence  |
| `score`  | Ordered list of level descriptions, 2–20 levels           | Expected level, per-level probabilities, legend |
| `noul`   | Optional descriptions for `false` and `true`              | Probability that the statement holds (0–1)      |

Every request can mix all three types, and every question is answered in the same forward pass.

## Why it matters

* **Structured:** replies are typed JSON values with probabilities, never free text.
* **Fast:** answers in tens of milliseconds per request on the hosted deployment.
* **Consistent:** the same state and questions return the same decision.
* **Side-effect free:** a classification call never charges, refunds, or retries anything.

## Next steps

* [Classify endpoint](/api-reference/beep/classify) — request format, examples, responses, and errors.