LLM-Powered Auto-Extract Fields

Overview

We’ve introduced a new LLM-Powered Auto-Extract Fields feature to simplify custom document type setup in Docsumo. This feature automates the extraction of field labels and table headers when users upload a document, significantly reducing manual effort and enhancing onboarding efficiency.

Feature Highlights

Automatic Field Suggestions: Extract field labels and table headers from uploaded documents.
Edit and Finalize: Users can review and adjust the suggested fields before finalizing. (Learn more about Edit Fields)
Improved User Experience: Streamlined setup process with minimal manual intervention.

How to Use the Auto-Extract Fields Feature

Step 1: Create a New Document Type:

Click on the Create Custom Document Type button.
Upload a sample document to initiate the setup and click "Setup Fields".

Step 2: Auto-Extract Fields:

The LLM model analyzes the document and displays suggested field labels and table headers.

Step 3: Review Suggestions:

Review and Edit the list of fields or add additional fields as necessary. Learn more.

Step 4: Finalize the Setup:

Once satisfied, click on "Save & Extract" to save the setup and complete the configuration for the document type.

FAQs

What happens if the LLM fails to extract fields?
If the extraction fails, the existing list of fields will be retained. If no fields are extracted, the user must manually add/edit the list of required fields.
Will the auto-extract feature work with non-standard document formats?
Yes, the LLM model is trained on various document types, but users can manually override and edit suggestions if needed.
Can I use the feature for existing document types?
This feature is only available for custom document types.

Should you have any questions or encounter any issues during the process, feel free reach out to us at [email protected], and we'll be more than happy to help you.