> For the complete documentation index, see [llms.txt](https://docs.agenticflow.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.agenticflow.ai/reference/nodes/extract_content.md).
# Extract Content
**Action ID:** `extract_content`
## Description
Extract structured content from text using a specified schema.
## Input Parameters
| Name | Type | Required | Default | Description |
| --------------- | -------- | :------: | ------------------ | -------------------------------------------------------------------------------------------------- |
| extract\_from | string | ✓ | - | The content to extract from |
| extract\_schema | object | ✓ | - | A JSON schema representing the structure for extracted content |
| model | dropdown | - | gpt-3.5-turbo-0613 | The LLM model to use for extracting content. Available options: gpt-3.5-turbo-0613, gpt-4-32k-0613 |
View JSON Schema
**Input Schema**
```json
{
"$defs": {
"AllowedModels": {
"description": "Allowed models for the web scraping node.",
"enum": [
"gpt-3.5-turbo-0613",
"gpt-4-32k-0613"
],
"title": "AllowedModels",
"type": "string"
}
},
"description": "Extract content node input.",
"properties": {
"extract_from": {
"description": "The content to extract from.",
"title": "Extract From",
"type": "string"
},
"extract_schema": {
"additionalProperties": true,
"description": "A json string represent schema for the extracted content.",
"title": "Extract Schema",
"type": "object"
},
"model": {
"$ref": "#/$defs/AllowedModels",
"default": "gpt-3.5-turbo-0613",
"description": "The LLM model to use for extracting content.",
"title": "Model"
}
},
"required": [
"extract_from",
"extract_schema"
],
"title": "ExtractContentNodeInput",
"type": "object"
}
```
## Output Parameters
| Name | Type | Description |
| ------------------ | ----- | --------------------------------------------------------------------------------------- |
| extracted\_content | array | The extracted content from the text as an array of objects matching the provided schema |
View JSON Schema
**Output Schema**
```json
{
"description": "Extract content node output.",
"properties": {
"extracted_content": {
"description": "The extracted content from the web URLs.",
"items": {
"additionalProperties": true,
"type": "object"
},
"title": "Extracted Content",
"type": "array"
}
},
"required": [
"extracted_content"
],
"title": "ExtractContentNodeOutput",
"type": "object"
}
```
## How It Works
This node uses large language models (LLMs) to intelligently parse unstructured text and extract specific information based on a JSON schema you define. The LLM analyzes the input text, identifies relevant data points matching your schema structure, and returns the extracted information in a structured, consistent format. This enables automated data extraction from emails, documents, web content, and other text sources.
## Usage Examples
### Example 1: Extract Contact Information
**Input:**
```
extract_from: "John Smith is the CEO. You can reach him at john@example.com or call (555) 123-4567. His office is at 123 Main St, New York, NY 10001."
extract_schema: {
"type": "object",
"properties": {
"name": {"type": "string"},
"email": {"type": "string"},
"phone": {"type": "string"},
"address": {"type": "string"}
}
}
model: "gpt-3.5-turbo-0613"
```
**Output:**
```
extracted_content: [
{
"name": "John Smith",
"email": "john@example.com",
"phone": "(555) 123-4567",
"address": "123 Main St, New York, NY 10001"
}
]
```
### Example 2: Extract Product Information
**Input:**
```
extract_from: "We have the iPhone 15 Pro available for $999. The Samsung Galaxy S24 costs $899 and comes in blue, black, and silver. The Google Pixel 8 is priced at $699."
extract_schema: {
"type": "object",
"properties": {
"product_name": {"type": "string"},
"price": {"type": "number"},
"colors": {"type": "array", "items": {"type": "string"}}
}
}
model: "gpt-4-32k-0613"
```
**Output:**
```
extracted_content: [
{
"product_name": "iPhone 15 Pro",
"price": 999,
"colors": []
},
{
"product_name": "Samsung Galaxy S24",
"price": 899,
"colors": ["blue", "black", "silver"]
},
{
"product_name": "Google Pixel 8",
"price": 699,
"colors": []
}
]
```
### Example 3: Extract Event Details
**Input:**
```
extract_from: "Join us for the Tech Summit 2024 on March 15-17 at the Convention Center. Registration opens at 8 AM. Keynote speaker: Dr. Jane Doe. Cost: $299 for early bird."
extract_schema: {
"type": "object",
"properties": {
"event_name": {"type": "string"},
"dates": {"type": "string"},
"location": {"type": "string"},
"speaker": {"type": "string"},
"price": {"type": "number"}
}
}
model: "gpt-3.5-turbo-0613"
```
**Output:**
```
extracted_content: [
{
"event_name": "Tech Summit 2024",
"dates": "March 15-17",
"location": "Convention Center",
"speaker": "Dr. Jane Doe",
"price": 299
}
]
```
## Common Use Cases
* **Email Processing**: Extract key information like names, dates, and action items from emails
* **Document Parsing**: Pull structured data from invoices, contracts, and business documents
* **Web Scraping**: Extract specific data points from web page content
* **Customer Data Extraction**: Parse customer inquiries to extract contact details and requirements
* **Product Catalog Creation**: Extract product details from descriptions to build structured catalogs
* **Resume Parsing**: Extract candidate information like skills, experience, and education from resumes
* **Lead Generation**: Extract business contact information from various text sources
## Error Handling
| Error Type | Cause | Solution |
| --------------------- | ----------------------------------------------- | ---------------------------------------------------------------------------------- |
| Invalid Schema | JSON schema is malformed or invalid | Validate your JSON schema structure and ensure it follows proper JSON syntax |
| Model Error | LLM API is unavailable or rate limited | Retry the operation or switch to an alternative model |
| Empty Content | extract\_from field is empty or null | Provide valid text content for extraction |
| Schema Mismatch | Content doesn't match expected schema structure | Adjust schema to match the actual content structure or provide appropriate content |
| Token Limit Exceeded | Input text is too long for the model | Split text into smaller chunks or use gpt-4-32k-0613 for larger content |
| Authentication Failed | Invalid API credentials | Verify your OpenAI API connection and credentials |
| Extraction Failed | LLM unable to extract matching data | Simplify the schema or provide more explicit content that matches the schema |
## Notes
* **Schema Design**: Design clear, specific schemas that match your expected data structure. Include field types and nested objects as needed.
* **Model Selection**: Use gpt-3.5-turbo-0613 for simple extractions and gpt-4-32k-0613 for complex content or higher accuracy requirements.
* **Content Quality**: Better structured input text produces more accurate extraction results.
* **Array Output**: The node returns an array of objects, allowing extraction of multiple items from a single text source.
* **Data Types**: Ensure your schema specifies appropriate data types (string, number, boolean, array, object) for accurate extraction.
* **Performance**: Extraction time varies based on content length and model choice. GPT-4 is slower but more accurate than GPT-3.5.