> ## 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

> The decision model behind Paytring AI: typed answers with probabilities.

**Beep** is Paytring's decision model. It reads a state — a ticket, an email, a checkout note, or any JSON document — and answers the typed questions you define. It is not a chatbot: it never writes prose. Every response is a typed decision with probabilities you can automate on.

## What it can do

<CardGroup cols={2}>
  <Card title="Answer in three decision types" icon="list-check">
    **Choice:** pick one of your labels, such as `billing` or `high_risk`. **Score:** place the state on a scale you define, such as urgency from 0 to 2. **Probability:** return how likely a statement is, from 0 to 1.
  </Card>

  <Card title="Return probabilities, not just a guess" icon="chart-simple">
    Every option comes with its probability, and each answer carries a confidence value. High-confidence answers can be automated; low-confidence ones can be routed to a human.
  </Card>

  <Card title="Answer several questions at once" icon="layer-group">
    Up to four typed questions about the same state are evaluated together, and every answer comes back in one response.
  </Card>

  <Card title="Read text or structured data" icon="brackets-curly">
    Plain sentences, support messages, order and payment JSON, or conversation turns. Beep reads the state and the questions together.
  </Card>

  <Card title="Use your own labels, no training" icon="tags">
    You define the labels, scales, and wording at request time. Adding a new category or queue never requires a new model.
  </Card>

  <Card title="Behave predictably in production" icon="bolt">
    The same state and questions produce the same decision. Responses arrive in tens of milliseconds and never trigger a payment, refund, or retry.
  </Card>
</CardGroup>

## Decisions Beep can express

| Question you might ask                       | Decision type | Example answer  |
| -------------------------------------------- | ------------- | --------------- |
| Which department should handle this request? | Choice        | `billing`       |
| How urgent is this request?                  | Score         | `1.96` of `2`   |
| Does the customer request a refund?          | Probability   | `0.81`          |
| Is this customer about to churn?             | Probability   | `0.74`          |
| Which payment method fits this checkout?     | Choice        | `card`          |
| Should this failed payment be retried?       | Choice        | `switch_method` |

The same model expresses all of these because the decision is defined by your questions, not baked into the model.

## Built for payment operations

Beep powers the workflows in [AI Use Cases](/ai/use-cases): transaction risk signals, payment method prediction, duplicate charge and refund triage, failed-payment retries, dispute routing, and support ticket triage.

<Note>
  Access is enabled per account, authenticated with **HTTP Basic** using your Paytring credentials. Contact your **Paytring account manager** to get started.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="AI Use Cases" icon="sparkles" href="/ai/use-cases">
    Payment workflows you can build with Beep.
  </Card>

  <Card title="Classify API" icon="code" href="/api-reference/beep/classify">
    Request format and response reference, when you are ready to integrate.
  </Card>
</CardGroup>