Entity-Model Automation for Predictable Distributed Agent Actions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI systems struggle to automate complex entities like organizations effectively due to their specific and dynamic nature, requiring highly customizable models that can adapt to varying structures and contexts, while current large models exhibit unpredictable behaviors.
Innovation Solution
A modular intelligence architecture with a central system coordinating autonomous agents and services, using explicit entity models and artefacts to enable predictable and scalable automation, allowing for dynamic context matching and agent engagement based on model updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If large, deep machine learning models are used for automation, then the system can perform a wide range of tasks, but the behavior becomes hard to predict
Solution Approach 1:
The system segments intelligence into multiple specialized autonomous agents, each with specific expertise and capabilities. Instead of relying on a single large model, the system divides functionality into discrete, controllable agents that can be independently managed and whose behaviors are more predictable within their domains.
Solution Approach 2:
A central system acts as an intermediary between autonomous agents and external services. This mediator coordinates agent actions, manages entity models, and provides a layer of control that enhances predictability while maintaining the versatility of multiple specialized agents.
2Adaptability or versatility
If a centralized system coordinates autonomous agents, then the system becomes highly scalable, but the device complexity increases
Solution Approach 1:
The central system implements a universal entity model that can represent diverse entities and relationships in a standardized way. This multi-functional model handles various types of data and coordination scenarios, reducing the need for specialized components and simplifying the overall architecture despite the system's scalability.
3Loss of information
If explicit entity models with artefacts are used, then the system provides explainable decision automation, but the manufacturing precision requirements increase
Solution Approach 1:
The system uses artefacts with defined parameters and properties that can be dynamically instantiated and configured. By changing parameters of existing artefact templates rather than creating entirely new models, the system maintains precision while improving efficiency and reducing the burden of model definition.
Data Source
AI summary
The present disclosure provides systems and methods performed in a central system. An initial entity model comprising a plurality of artefacts is formed. A runtime engine determines a plurality of available agent engagements based on the initial entity model. Responsive to an input selecting one of the available agent engagements, a notification is sent to an external autonomous agent. An indication of an action to be performed is received from the external autonomous agent. Responsive thereto, a message is sent to an external service indicating the action, thereby causing the action to be performed. The entity model is updated with a new artefact corresponding to the performed action, and a further plurality of agent actions is determined based on the updated entity model.