USPTO Patent Grant: ML System for Document Processing

ChangeBridge: Patent Grants - AI & Computing (G06N)

Published March 24th, 2026

Detected March 24th, 2026

Summary

The USPTO has granted patent US12585706B2 for a machine-learning based system and method for automatically processing documents. The patent, assigned to HIGHRADIUS TECHNOLOGIES PRIVATE LIMITED, details a process for classifying document contents using ML models.

View original document View source feed page

What changed

The United States Patent and Trademark Office (USPTO) has granted patent US12585706B2 to HIGHRADIUS TECHNOLOGIES PRIVATE LIMITED. This patent covers a machine-learning based system and method for automatically processing documents, including classifying first and second contents as claim-based or non-claim-based using ML models and prompts. The patent was filed on December 30, 2024, and is set to be granted on March 24, 2026.

This is a patent grant, not a regulatory rule or guidance. As such, it does not impose new compliance obligations on regulated entities. However, companies developing or utilizing AI-powered document processing systems should be aware of this granted patent, particularly concerning its claims related to ML classification and user interface outputs. No immediate compliance actions are required.

Source document (simplified)

← USPTO Patent Grants

Machine-learning based (ML-based) system and method for automatically processing one or more documents

Grant US12585706B2 Kind: B2 Mar 24, 2026

Assignee

HIGHRADIUS TECHNOLOGIES PRIVATE LIMITED

Inventors

Parsuram Panigrahi, Somesh Manna, Chirodeep Kar, Susanta Kumar Sahoo, Nilesh Kumar, Atanu Mazumdar

Abstract

A machine-learning based (ML-based) system and method for automatically processing one or more documents are disclosed. Initially, data associated with at least one of: first contents and second contents are obtained from data sources. The data are pre-processed to generate pre-processed data. The first contents are classified are classified as at least one of: claim-based contents and non-claim-based contents, using a machine learning (ML) model with prompts. Additionally, the second contents are classified as at least one of: the claim-based contents and the non-claim-based contents, based on intents of the second contents, using the ML model with the prompts. At least one of: the claim-based contents and the non-claim-based contents, are provided as an output, to users on user interfaces associated with electronic devices associated with the users.