Trending Topics Onderzoeksassistent
Ontdek waar mensen online over praten rondom jouw gekozen trefwoord. Deze AI-gestuurde workflow onderzoekt trending of gerelateerde onderwerpen uit recente inte...
4 min lezen
The URL Retriever is a versatile flow component designed to fetch and process web content from specified URLs, returning the information as structured documents. It serves as a bridge between external online content and your AI workflow, enabling you to integrate, analyze, or process web-based information efficiently.
What Does It Do?
This component retrieves the content of one or multiple URLs provided as input. It can extract the main text, metadata, and even process content from images using Optical Character Recognition (OCR). The retrieved data is then made available in various structured formats suitable for downstream AI tasks such as summarization, question answering, or knowledge extraction.
Input Options
You can supply URLs to the component in two ways:
Text URLs:
- Input Type:
Message - Description: A list of plain URL links for the component to fetch content from.
- Input Type:
URL Records:
- Input Type:
UrlRecord - Description: A list of structured URL records, which may include additional metadata.
- Input Type:
Advanced Input Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| Apply OCR | Boolean | false | If enabled, applies OCR to extract text from images in the document. |
| Cache TTL | Dropdown | 2 weeks | How long the content should be cached, with options from no cache up to 1 year. |
| From H1 if exists | Boolean | true | Begins extraction from the H1 tag if present, focusing on main content. |
| Load from pointer | Boolean | true | Loads content starting from the most relevant section based on your query. |
| Hide Resources | Boolean | false | Hides the retrieved resources from being output or displayed. |
| Max Tokens | Integer | 3000 | Sets the maximum number of tokens for the output text. |
| Skip Last Header | Boolean | true | Skips the last header during extraction for streamlined content. |
| Strategy | Dropdown | Include equal size from each documents | Determines how content is combined: concatenate fully or include equal parts from each document. |
| Export Content | Multi-select | All | Choose which HTML elements to export (H1-H6, Paragraph). |
| Include Metadata | Multi-select | Product | Specify which metadata fields to include (e.g., Product, Author, Website, etc.). |
| Verbose | Boolean | false | Enables detailed output for debugging or information purposes. |
| Tool Name | String | (empty) | Optionally assign a custom name to the tool for agent reference. |
| Tool Description | Multiline | (empty) | Provide a description to help agents understand the tool’s purpose. |
Outputs
The URL Retriever provides its outputs in several formats, allowing flexible integration with various AI processes:
| Output Name | Type | Description |
|---|---|---|
| Documents | Message | The processed content from the URLs, ready for use in messaging-oriented workflows. |
| Raw Documents | Document | The raw, unprocessed document objects for advanced downstream processing. |
| Documents As Tool | Tool | The content packaged as a tool, enabling agent-based workflows to utilize the documents. |
Why Use the URL Retriever?
- Integrate External Knowledge: Seamlessly bring web-based information into your AI applications, such as chatbots, search engines, or knowledge bases.
- Customizable Extraction: Fine-tune what content and metadata you want, control the amount of data, and use OCR for images.
- Performance & Efficiency: Use caching to avoid redundant downloads, and limit token output for performance.
- Flexible Output Formats: Choose the output format that best fits your next workflow step—structured document, message, or tool.
Example Use Cases
- Building knowledge-grounded conversational agents that answer questions using up-to-date web content.
- Aggregating product data from e-commerce sites for comparison or analytics.
- Monitoring and analyzing blog or news articles based on specific topics or keywords.
- Extracting information from web pages containing mixed media (text and images).
Summary Table
| Feature | Description |
|---|---|
| Fetches URLs | Retrieves and processes web content from provided URLs. |
| OCR Support | Extracts text from images in documents if enabled. |
| Metadata Extraction | Optionally includes metadata such as author, product, or schema.org types. |
| Customizable Output | Select which HTML elements or metadata to export. |
| Caching | Configurable cache lifetimes for efficiency. |
| Multiple Output Types | Supports message, raw document, and tool outputs for workflow flexibility. |
The URL Retriever is a powerful and flexible bridge between web content and your AI workflows, offering granular control over content extraction and integration.
