# TypeSafe

> Connect TypeSafe so your workflows can make judgment calls: deciding what something is, not just what it says.

App · AI Models · 1 action. Web page: https://docs.lodol.com/docs/integrations/typesafe

## Actions

### Decide yes or no

Answer a yes/no question about something, with how likely the answer is to be yes.

**Inputs**

- Question (`question`, text, required): The yes/no question to answer, in your own words. For example: 'Is this email asking for a refund?'
- What to read (`about`, text): The text or data the question is about: an email body, a support ticket, a row from a sheet.
- Answer yes above (`threshold`, number): How likely it has to be before the answer reads 'yes'. A number from 0 to 1. Defaults to 0.5. Default: `0.5`.

**Outputs**

- `answer` (text)
- `model` (text)
- `probability` (number)
- `question` (text)
- `status` (text)

**Example output**

```json
{
  "answer": "yes",
  "model": "jev-latest",
  "probability": 0.91,
  "question": "Is this email asking for a refund?",
  "status": "success"
}
```