ML Workflow Step Prediction via Natural Language

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Developers face challenges in creating computerized workflows due to the complexity of graphical user interfaces and the need to learn coding languages, which can lead to inefficient workflow generation and a lack of adherence to best practices.

Innovation Solution

A machine learning-based system that predicts and implements additional steps in a computerized workflow using natural language inputs, allowing users to generate workflows without requiring knowledge of executable coding languages, by employing a text-to-text model that converts user descriptions into API calls for automation flow builders.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If developers use graphical user interfaces to create computerized workflows, then workflow generation becomes more accessible, but the complexity of learning the development tool increases

Engineering Contradiction:
Improveworkflow generation accessibilityVSAvoiddevelopment tool complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary system that translates natural language descriptions into executable workflow steps. This intermediary layer allows users to interact with complex workflow systems using simple natural language, without needing to learn the underlying graphical interface or coding language, thus resolving the contradiction between accessibility and tool complexity

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the mechanical interaction with graphical user interfaces and visual drag-and-drop operations with a language-based interaction model. By substituting the visual/mechanical interface with natural language processing, users can create workflows through text descriptions rather than manipulating complex graphical elements, reducing the learning curve while maintaining accessibility

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If developers learn coding languages to create workflows, then workflow functionality increases, but training time and learning curve increase

Engineering Contradiction:
Improveworkflow functionalityVSAvoidtraining time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent extracts the coding language requirement from the workflow creation process. By separating the natural language input from the executable workflow generation, the system eliminates the need for users to learn programming languages while still producing functional workflows, thus maintaining versatility without the time cost of training

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the input parameter from coded commands to natural language descriptions. This parameter change allows the system to interpret high-level intent descriptions and automatically generate the detailed workflow steps, maintaining full workflow functionality while eliminating the need for users to learn complex syntax and programming concepts

Inventive Principle:
Principle #35Parameter changes

3Manufacturing precision

If developers manually create each workflow step, then precision and control increase, but productivity decreases

Engineering Contradiction:
Improveworkflow step precisionVSAvoidworkflow generation speed
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The patent performs preliminary action by automatically generating complete workflow sequences from high-level natural language descriptions. The system pre-computes the necessary workflow steps, intermediate actions, and edge case handling before execution, allowing rapid workflow creation while maintaining precision through automated validation and best practice enforcement

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements self-service by enabling the workflow system to automatically generate, validate, and optimize workflow steps without manual intervention. The system serves itself by interpreting natural language input and autonomously creating precise workflow implementations, significantly increasing productivity while maintaining quality through built-in validation mechanisms

Inventive Principle:
Principle #25Self-service

4Ease of operation

If inexperienced users create workflows without guidance, then ease of use increases, but workflow quality and adherence to best practices decrease

Engineering Contradiction:
Improveuser accessibilityVSAvoidworkflow quality
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent incorporates feedback mechanisms that automatically analyze generated workflows for quality and best practice adherence. The system provides real-time feedback on workflow correctness, suggests improvements, and validates against established patterns, ensuring high-quality workflows are produced even by inexperienced users while maintaining ease of use

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent applies beforehand cushioning by pre-establishing best practice templates and validation rules that automatically guide workflow generation. These pre-configured quality safeguards cushion against poor workflow design by inexperienced users, ensuring reliability without adding complexity to the user interface or reducing accessibility

Inventive Principle:
Principle #11Beforehand cushioning (Prior cushioning)

Data Source

PatentEP4283546A1Machine learning prediction of additional steps of a computerized workflow
Publication Date: 2023.11.29 SERVICENOW INC
  • EP4283546A1 patent drawingFigure 1
  • EP4283546A1 patent drawingFigure 1A
  • EP4283546A1 patent drawingFigure 2

AI summary

An indication to predict one or more additional steps to be added to a partially specified computerized workflow based at least in part on the partially specified computerized workflow is received. Text descriptive of at least a portion of the partially specified computerized workflow is generated. Machine learning inputs based at least in part on the descriptive text are provided to a machine learning model to determine an output text descriptive of the one or more additional steps to be added. One or more processors are used to automatically implement the one or more additional steps to be added to the partially specified computerized workflow.