LLM Agent Orchestration for Codeless Procurement Data Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing enterprise applications developed by codeless platforms face challenges in real-time data processing, data abstraction, and deriving meaningful insights due to varied data formats and complex interactions, with existing large language models (LLM) struggling to scale effectively for multiple tasks.
Innovation Solution
A system and method utilizing a layered architecture with Master and micro LLM agents, a tool repository, and a tool execution agent to process inputs, enabling efficient execution of tasks in procurement and supply chain applications through a codeless platform, with fine-tuned LLM agents and semantic search capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing large language models are used for data processing in enterprise applications, then text data processing capability is provided, but scalability for multiple tasks and varied data formats is limited
Solution Approach 1:
The system segments the monolithic LLM into multiple specialized micro-LLM agents, each trained on specific data formats and tasks. This segmentation allows the system to handle varied data formats (structured, unstructured, semi-structured) through specialized agents while maintaining overall system coordination through a master controller, thereby improving adaptability without proportionally increasing complexity.
Solution Approach 2:
The master controller LLM agent serves as a universal coordinator that can route any type of request to appropriate micro-LLM agents. This multi-functional component handles diverse data formats and task types through a single interface, improving versatility while managing complexity through centralized control logic.
2Productivity
If multiple LLM agents are deployed to handle different tasks, then task execution capability is improved, but scalability and coordination complexity increases
Solution Approach 1:
The master controller LLM agent acts as an intermediary between user requests and multiple micro-LLM agents. It receives natural language queries, determines the appropriate specialized agents needed, and coordinates their execution. This intermediary layer improves productivity by enabling parallel processing of multiple tasks while managing coordination complexity through a single point of control.
Solution Approach 2:
Instead of having a single LLM handle all tasks directly, the system inverts the architecture by having multiple specialized LLM agents handle specific tasks and a master controller orchestrate them. This inversion allows distributed task execution (improving productivity) while centralizing coordination logic (managing complexity).
3Speed
If real-time data processing is implemented in enterprise applications, then operational responsiveness is improved, but data processing complexity and resource requirements increase
Solution Approach 1:
The system applies local quality by training each micro-LLM agent on specific data formats and task types relevant to their function. This specialization allows each agent to process its designated data type efficiently in real-time without the overhead of a general-purpose model, improving processing speed while managing complexity through targeted specialization.
4Ease of manufacture
If codeless platform is used for application development, then ease of development is improved, but support for data abstraction and real-time modifications is limited
Solution Approach 1:
The LLM-based system provides self-service capabilities by automatically understanding natural language requests and translating them into appropriate data processing operations. This eliminates the need for users to manually configure complex data abstraction layers, maintaining ease of use while enabling sophisticated data handling through the specialized micro-LLM agents.
Data Source
AI summary
The present invention provides a system and method for data processing in procurement and supply chain application developed by codeless platform. The invention includes one or more large language model (LLM) agents configured for processing one or more input received on an electronic user interface. The invention includes selecting a tool selection agent from a tool repository and invoking the selected tool by a tool execution agent for processing at least one task to be executed.