Ekstrakcja danych z e-maili i plików do CSV
Ten przepływ pracy wyodrębnia i porządkuje kluczowe informacje z e-maili oraz załączonych plików, wykorzystuje AI do przetwarzania i strukturyzacji danych, a na...
4 min czytania
The Structured Generator component is designed to generate structured data outputs by leveraging Large Language Models (LLMs). This component is particularly useful when you need AI-generated text or data to conform to a specific schema or structure, making it easier to integrate with downstream systems, analytics, or workflows.
What Does This Component Do?
The Structured Generator takes user input in the form of a prompt, optionally incorporates images and system-level instructions, and employs a selected LLM to produce outputs that follow a user-defined structure. This makes it ideal for scenarios where consistent and predictable data formatting is required—such as form filling, report generation, or structured data extraction from text.
Inputs
The component supports several configurable inputs:
| Input Name | Type | Required | Description |
|---|---|---|---|
| Input | Message (Text) | No | The main prompt or question you want the LLM to process. |
| LLM | BaseChatModel | No | The language model instance used to generate the output. |
| Images | FlowSessionAttachmentResponse | No | Images to provide as additional context to the LLM. |
| Output Structure | NestedDict | Yes | Defines the desired structure and fields of the output. (e.g., specify keys and data types) |
| System Message | String (Multiline, Optional) | No | Additional instructions or context for the LLM at the system level (advanced, optional). |
Input Details
- Input (Prompt): The main instruction or question to be addressed by the LLM.
- LLM (Model): Select which language model to use for generating the response.
- Images: Attach images if the LLM supports multi-modal input and you want visual context.
- Output Structure: Specify the output format, such as the fields and their descriptions, ensuring the generated content is structured accordingly.
- System Message: (Advanced) Provide additional context or rules to guide the model’s behavior.
Output
The component produces a single output:
| Output Name | Type | Description |
|---|---|---|
| Structured Output | Data | The generated data, formatted as specified by the Output Structure field. |
This output can then be consumed by other components or exported for further processing.
Why Use the Structured Generator?
- Consistency: Ensures that the AI-generated outputs always adhere to a predefined schema, reducing post-processing effort.
- Flexibility: Supports a wide range of output structures, from simple fielded text to complex nested data.
- Multi-modal Input: Optionally include images alongside text to enrich the prompt for models that support it.
- Customizability: Use system messages to fine-tune the model’s behavior and output style according to your workflow needs.
- Integration: Ideal for AI workflows where structured data is required for API responses, databases, or analytics pipelines.
Example Use Cases
- Extracting structured information (like names, dates, and locations) from unstructured text.
- Generating standardized reports or summaries from user inputs.
- Filling forms or tables based on free-form descriptions.
- Producing JSON outputs for use in web applications or downstream processing.
Summary
The Structured Generator is a powerful and flexible tool for generating AI-driven structured data outputs. By configuring its input fields and output schema, you can tailor it to a wide variety of applications where structured, reliable, and repeatable data generation is essential.
