AI Workflow Automation Using Segmented LLM Object-Path Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing no/low code workflow builders are inefficient for generating complex workflows with conditional steps and multiple dependencies, requiring users to navigate hierarchical trees and identify data paths, which is cumbersome for non-coding users.
Innovation Solution
An AI workflow assistant uses a suite of large language models to analyze user input and generate machine-readable workflow descriptions, bifurcating the workload among two LLMs to identify objects and data paths, and either generate or transmit instructions for code creation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If users manually navigate hierarchical trees to identify data paths in no/low code workflow builders, then users can construct workflows, but the process becomes cumbersome and inefficient for complex workflows with conditional steps and multiple dependencies
Solution Approach 1:
The system enables self-service by allowing the LLM to automatically identify objects and data paths without user intervention. The LLM processes natural language descriptions and autonomously navigates the hierarchical data structures to locate required objects and determine optimal data paths, eliminating the manual navigation burden from users.
Solution Approach 2:
The LLM acts as an intermediary between the user's natural language description and the workflow construction system. It translates high-level user intent into specific object identifiers and data path selections, bridging the gap between simple user input and the complex hierarchical structure of the workflow builder.
2Productivity
If the system automates workflow construction using LLMs, then workflow generation accuracy and efficiency improve, but the system complexity increases due to multiple LLMs and bifurcated workload
Solution Approach 1:
The system segments the workflow construction task into two distinct sub-tasks handled by separate LLMs: one LLM identifies objects and their properties, while the other LLM determines data paths between objects. This segmentation allows each LLM to specialize in a specific aspect of workflow construction, improving overall accuracy while distributing system complexity across multiple independent components.
Data Source
AI summary
Disclosed herein are methods and systems to generate code for a workflow. A non-limiting example of a method comprises executing, by a processor, a large language model, which receives an input of an intent associated with a workflow and provides a machine-readable description by: identifying, by searching a hierarchical data structure using a vector embedding associated with the intent, an object corresponding to the intent; identifying a set of data paths within the hierarchical data structure to retrieve the identified object; and generating the machine-readable description, the machine-readable description describing the object and at least one data path within the set of data paths; and generating, by the processor, code for the workflow using the machine-readable description.