Chapter 21 Configuring Inference Agents (Cache OM) : Inference Agent Runtime Architecture

Inference Agent Runtime Architecture
The diagram below shows how messages come into an inference agent, and are processed through the system, as explained below. The section Multiple Agent Write and Replication Step explains how changes in one agent’s working memory are communicated to all instances of the agent.
Single Agent Flow
Each event entering a channel is handled by a worker thread in the thread pool You can define a number of worker threads (see Input Destinations). The flow is as follows:
1.
2.
3.
4.
5.