LLM Agent Selection for Adaptive Workflow Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing next best agent recommendation systems for adaptive workflows rely on historical data and deterministic rules, which do not scale or adapt to changing workflows, new agents, and do not conform to API data requirements, making them ineffective in dynamic environments.
Innovation Solution
A method that extracts goals from user utterances, uses a large language model (LLM) to select candidate agents based on metadata and business policies, scores them, and invokes the next best agent to further the goal, without relying on historical data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If historical data and deterministic rules are used for agent recommendation, then the system is simple to implement, but it cannot adapt to changing workflows and new agents
Solution Approach 1:
The patent replaces deterministic mechanical rules with a large language model-based system that uses natural language processing and contextual understanding to recommend agents. This substitution enables the system to adapt to changing workflows and new agents without requiring manual reconfiguration of rigid rules.
Solution Approach 2:
The system dynamically changes its behavior based on contextual parameters such as current workflow state, agent metadata, and business policies. By adjusting recommendations based on these varying parameters, the system maintains adaptability while managing complexity through parameter-driven decision-making.
2Manufacturing precision
If deterministic rules are used for agent selection, then the system is easy to operate, but it does not conform to API data requirements
Solution Approach 1:
The large language model automatically analyzes API data requirements and business policies to generate compliant agent recommendations without requiring manual verification. The system self-adjusts to ensure data conformance while maintaining ease of operation through automated decision-making.
3Adaptability or versatility
If historical data is relied upon for recommendations, then the system is simple to implement, but it cannot scale to new agents and workflows
Solution Approach 1:
The system performs preliminary analysis of agent metadata and business policies before generating recommendations. By preparing and analyzing this information in advance, the system can scale to new agents and workflows without requiring historical data, as the preliminary processing enables immediate adaptation to new elements.
Data Source
AI summary
An embodiment extracts, from an utterance, a goal. An embodiment prompts a large language model (LLM) to select, using metadata describing a plurality of agents, a set of candidate agents from the plurality of agents, each candidate agent in the set of candidate agents corresponding to the goal. An embodiment scores, using metadata of the set of candidate agents, each candidate agent in the set of candidate agents, the scoring resulting in a set of scored candidate agents. An embodiment prompts the LLM to select, using a set of business policy constraints, a next agent from the set of scored candidate agents. An embodiment invokes the next agent, the invoking causing the next agent to perform an action furthering the goal.