# Introduction

Waveline Extract is a powerful AI-driven service designed for developers to effortlessly extract information from unstructured data using our API.

We currently support two services:

* [Extract-Document](https://docs.waveline.ai/extract/endpoints/extract-document) (Extract specific JSON information)
* [Guess-Shape](https://docs.waveline.ai/extract/endpoints/guess-shape) (Guess what Information you could extract)

## Example (extract-document)

You provide us with a **Document** (this can be plain text, a PDF, ...) and a [**Shape**](https://docs.waveline.ai/extract/types/shape) (describes what information you want to extract), and we will return you a **JSON** with that information extracted from your document. 

**Document**: 

```
Dear Resort Aroma,

I would like to book a room from the 25th of April until the 30th of April.
When I looked through the website, I saw that you have rooms with a balcony. 
It would be wonderful if I could get such a room.  

Best,
Richard Fulmar
```

**Shape**:

<pre class="language-json"><code class="lang-json"><strong>"shape": [
</strong>  {
    "name": "name",
    "type": "string",
    "description": "First and last name of the guest",
    "isArray": false
  }
]
</code></pre>

After sending us these two things, we return you a JSON containing the desired information.

**Returned JSON:**

```json
{
  "name": "Richard Fulmar"
}
```

## Example (guess-shape)

You provide us with a **Document** (this can be plain text, a PDF, ...). We will then guess the most important types of information in it and return it in the form of a [**Shape**](https://docs.waveline.ai/extract/types/shape)**.**

**Document**:

```
Dear Resort Aroma,

I would like to book a room from the 25th of April until the 30th of April.
When I looked through the website, I saw that you have rooms with a balcony. 
It would be wonderful if I could get such a room.  

Best,
Richard Fulmar
```

We then return a shape with fields we think you want to extract.

**Returned Shape:**

```json
"shape": [
  {
    "name": "name",
    "type": "string",
    "description": "First and last name of the guest",
    "isArray": false
  },
  {
    "name": "room_preference",
    "type": "string",
    "description": "Guest's preferred room location, room size, bed type,...",
    "isArray": false
  },
  {
    "name": "reservation_number",
    "type": "number",
    "description": "Unique number that identifies this reservation",
    "isArray": false
  }
]
```

## How does it work?

The core of our processing is done with Large Language Models (LLMs) like GPT-4, Claude, PaLM, etc. The difference to conventional methods like regex-parsing and keyword-matching is that we are much more flexible and can handle more complicated queries this way. While LLMs have their own flaws, like hallucinations, incorrect formatting, and slow performance, we take care of that and only give the advantages of LLMs to our end-users.\
\
Under to hood, we give the name, type, and description of each shape element to the Language Model. When using our API, you can help us by providing descriptive names and accurate descriptions for better performance.&#x20;


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.waveline.ai/extract/waveline-extract/introduction.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.