Machine-Learning Workflow Generation from Natural-Language Input

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Developers face challenges in creating computer instructions due to the complexity of learning to use development tools and the difficulty of following best practices, especially when using graphical user interfaces for workflow automation.

Innovation Solution

A system that uses a machine learning model to automatically generate computerized workflows from natural language descriptions, allowing users to input workflows using free-form text and leveraging large pre-trained language models to predict actions and convert them into API calls, reducing the need for manual processing and feature engineering.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If developers use graphical user interfaces for workflow automation, then they can create computer instructions without significant knowledge of computer language, but they face difficulty learning to use the development tool and are overwhelmed by many options

Engineering Contradiction:

Improveease of creating computer instructionsVSAvoidease of using development tool

Core Design Contradiction:

Ease of manufactureVSEase of operation

Solution Approach 1:

The patent introduces a natural language interface as an intermediary between the user and the complex graphical user interface. Users can describe workflows in plain English, and the system automatically translates these descriptions into structured workflows, eliminating the need for users to navigate complex GUI options while still providing access to sophisticated automation capabilities.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the mechanical interaction with graphical UI elements (clicking, dragging, configuring) with a natural language processing system. Instead of manually configuring workflow parameters through GUI controls, users simply type their intentions in natural language, and the system automatically generates the corresponding workflow structure.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If developers manually configure workflow parameters through graphical interface, then they can control workflow details, but they may not utilize best practices and spend excessive time on manual processing

Engineering Contradiction:

Improvecontrol over workflow detailsVSAvoidworkflow generation efficiency

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The system performs preliminary analysis of the natural language input to automatically determine appropriate workflow parameters, data transformations, and configuration settings. This preliminary processing eliminates the need for users to manually configure each parameter while ensuring that best practices are automatically applied based on the system's pre-programmed knowledge.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system automatically generates complete workflow configurations from natural language descriptions without requiring manual intervention. The workflow engine self-services by interpreting user intent, selecting appropriate parameters, and constructing the workflow structure autonomously, thereby maintaining precision while dramatically improving productivity.

Inventive Principle:

Principle #25Self-service

3Ease of operation

If system automatically generates workflows from natural language, then it reduces learning curve and promotes best practices, but it requires sophisticated machine learning model and processing

Engineering Contradiction:

Improvelearning curve for usersVSAvoidcomplexity of machine learning system

Core Design Contradiction:

Ease of operationVSDevice complexity

Solution Approach 1:

The patent employs a large pre-trained language model that serves multiple functions: understanding natural language input, extracting workflow parameters, determining data transformations, and generating workflow structure. This universal model eliminates the need for separate specialized systems for each task, reducing overall system complexity while maintaining sophisticated capabilities.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system leverages the adaptive parameters and fine-tuning capabilities of pre-trained language models to handle diverse workflow scenarios. By adjusting model parameters and using few-shot learning techniques, the system can adapt to different workflow types without requiring fundamentally different processing architectures, thereby managing complexity while providing ease of use.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250315215A1Automatic flow implementation from text input

Publication Date: 2025.10.09 SERVICENOW INC

AI summary

A user provided text description of at least a portion of a desired workflow is received. Context information associated with the desired workflow is determined. Machine learning inputs based at least in part on the text description and the context information are provided to a machine learning model to determine an implementation prediction for the desired workflow. One or more processors are used to automatically implement the implementation prediction as a computerized workflow implementation of at least a portion of the desired workflow.