Prompt Orchestration Segregating Tasks for Parallel Processing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing artificial intelligence-based prompt orchestration systems face challenges due to the high computational resources required for training large neural networks, leading to prolonged training times and resource bottlenecks, which hinder efficient prompt generation.

Innovation Solution

The system segregates processing requests into non-serialized, function-specific subtasks, allowing for parallel processing without initially using the artificial intelligence model. This includes preprocessing data for functions that don't require AI, creating an aggregation function to prepare results for AI input, and optimizing resource allocation by using less computationally demanding resources for less complex functions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If artificial intelligence models are used for prompt orchestration, then flexibility and adaptability are improved, but computational resource requirements and processing time increase significantly

Engineering Contradiction:

Improveprompt orchestration flexibilityVSAvoidcomputational resource consumption

Core Design Contradiction:

Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The patent segments the prompt orchestration process into multiple independent components: rule-based processing for deterministic tasks, AI model processing for complex ambiguous tasks, and task routing logic. This segmentation allows the system to use computational resources selectively rather than always invoking AI models, reducing overall resource consumption while maintaining flexibility where needed.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different processing qualities to different tasks: simple, structured tasks receive rule-based processing with high efficiency, while complex, unstructured tasks receive AI-based processing with high adaptability. This local differentiation optimizes resource allocation by matching processing capability to task requirements rather than using a uniform approach.

Inventive Principle:

Principle #3Local quality

2Adaptability or versatility

If artificial intelligence models are used for prompt orchestration, then adaptability is improved, but processing speed and throughput decrease

Engineering Contradiction:

Improveprompt orchestration flexibilityVSAvoidprompt generation throughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments tasks based on complexity and routing them to appropriate processing pipelines. Rule-based tasks are processed quickly through deterministic logic, while only complex tasks requiring interpretation are routed to AI models. This segmentation maintains high throughput for the majority of tasks while preserving adaptability for complex cases.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies AI model processing partially - only to the subset of tasks that require it - rather than applying it universally. This partial action approach maintains productivity by avoiding the overhead of AI processing for tasks that don't require it, while still providing adaptability where needed.

Inventive Principle:

Principle #16Partial or excessive action

3Adaptability or versatility

If multiple artificial intelligence models are added to handle different functions, then versatility is improved, but computational resource requirements increase exponentially

Engineering Contradiction:

Improvefunction coverageVSAvoidcomputational resources

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent implements a universal task routing system that can direct different types of tasks to the most appropriate processing method (rules or AI). This universal approach replaces the need for multiple specialized AI models, as a single AI model can handle various complex tasks through the standardized routing interface, reducing resource requirements while maintaining versatility.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent uses a single AI model instance that can be reused across multiple function calls through the task routing system, rather than deploying multiple separate AI model instances. This copying approach allows one model to serve multiple purposes, significantly reducing the computational resources needed compared to having separate models for each function.

Inventive Principle:

Principle #26Copying

4Device complexity

If sequential processing is used with artificial intelligence models, then resource allocation is simplified, but processing time and user experience deteriorate

Engineering Contradiction:

Improveresource management simplicityVSAvoidprompt generation delay

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

The patent segments the processing pipeline into parallel stages: task analysis and routing occur simultaneously with preliminary data preparation, while rule-based and AI-based processing occur in parallel for different task subsets. This segmentation enables concurrent execution of independent operations, reducing overall processing time without significantly increasing resource management complexity.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250190714A1Systems and methods for implementing an artificial intelligence-based solution for prompt orchestration by segregating processing requests into non-serialized tasks

Publication Date: 2025.06.12 CITIBANK N A

AI summary

Systems and methods are described herein for a prompt engine microservice. The system may segregate a received request into a search function, a calculation function, and a schema selection function. The system may also create a new aggregation function that aggregates the results of the various processes into an input for an artificial intelligence model. By doing so, the system may process the new plurality of tasks in parallel and without the initial use of the artificial intelligence model.