LLM Task Automation System for RPA Script Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current RPA solutions require expertise and manual modifications for task automation, especially for tasks involving multiple screen transitions, and they face limitations in handling large amounts of information and securing personal data.

Innovation Solution

A method and system utilizing a large-scale language model (LLM) to automate tasks by receiving task descriptions, generating object information lists, selecting candidate objects, and creating automation scenarios, while ensuring security by modifying sensitive information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional RPA solutions are used, then task automation can be achieved, but users require IT expertise and manual script modifications are needed

Engineering Contradiction:

Improveease of task automationVSAvoidcomplexity of automation process

Core Design Contradiction:

Ease of operationVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary system comprising an object recognition module, information extraction module, and RPA script generation module that mediates between the user's natural language description and the automated execution. This intermediary automatically converts high-level task descriptions into executable RPA scripts, eliminating the need for users to manually write or modify complex automation code while maintaining task automation capability

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system enables self-service automation by allowing users to describe tasks in natural language without requiring IT expertise. The automated system independently performs object recognition, information extraction, script generation, and execution, making the automation process accessible to end users who can define their own tasks without professional developer intervention

Inventive Principle:

Principle #25Self-service

2Extent of automation

If RPA processes are automatically generated through generative AI, then task flow design is automated, but additional manual modifications are still required

Engineering Contradiction:

Improveautomation of task flow designVSAvoidtime for manual modifications

Core Design Contradiction:

Extent of automationVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by automatically generating complete, executable RPA scripts including object recognition logic, information extraction rules, and control flow structures before user execution. The system pre-processes the task description to create a fully functional automation script that requires no manual modification, thereby eliminating the time loss associated with post-generation adjustments

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If tasks involving multiple screen transitions are automated, then complex workflows can be handled, but automatic control over target systems is difficult to achieve

Engineering Contradiction:

Improvecapability to handle multiple screen transitionsVSAvoidcomplexity of controlling target system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements feedback mechanisms where the object recognition module continuously monitors the current screen state, compares it with expected states, and adjusts the automation flow accordingly. This feedback loop enables the system to automatically adapt to screen transitions and maintain proper control over target systems across multiple screens, handling complex workflows without increasing user-facing complexity

Inventive Principle:

Principle #23Feedback

4Loss of information

If large amounts of information are transmitted to external LLMs, then comprehensive task understanding is achieved, but personal information security risks increase

Engineering Contradiction:

Improvecompleteness of task informationVSAvoidsecurity risk of personal information

Core Design Contradiction:

Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent extracts and separates personal information from the task description before transmitting to external LLMs. The information extraction module identifies and removes sensitive data elements, processing only the necessary task-related information externally while keeping personal information secure within the local system, thus achieving comprehensive task understanding without exposing sensitive data

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20250173180A1Method and system for providing task automation service using llm

Publication Date: 2025.05.29 SAMSUNG SDS CO LTD

AI summary

A method for providing task automation service using LLM is provided. The method according to some embodiments may include receiving a description of a target task, generating a first object information list by collecting information on a plurality of objects displayed on a first execution screen of a program used for the target task, selecting at least one candidate object related to the target task from among the plurality of objects by feeding the first object information list and the description of the target task into the LLM, selecting a first target object from among the at least one candidate object by feeding information of the selected candidate object and the description of the target task into the LLM and generating an automation scenario corresponding to the target task based on a first activity related to the first target object.