A system for distributed AI multi-agent networks with fault isolation and operational maintenance based on the DALI building network architecture.

Abstract

The useful solution refers to a distributed workflow scheduling system with fault isolation and operation maintenance for multiple AI agents, inspired by the distributed principles of DALI networks. The system includes multiple AI agents, a semantic operational bus, a workflow coordinator, a fault isolator, and an operation maintenance controller. Each AI agent is configured to store local workflow state and related metadata to support workflow execution. The AI ​​agents exchange information through semantic operational signaling. The system allows the workflow to be divided into multiple fault isolation domains to limit fault propagation and maintain at least part of the workflow execution when a failure occurs.