Objective

The objective of this tutorial is to demonstrate how to create a Three Agent Collection scenario running under mimOE-ai. A Three Agent Collection, in this case, is a configuration of mimOE-ai that has a Coordinator Machine, two AMD/ARM Agent Machine and an Nvidia Orin Nano machine. The Coordinator Machine accepts a prompt from a client and then passes the prompt on the Three Agent Collection for processing.

Intended Reader

The intended reader for this tutorial is a developer who is familiar with mimik technology and the mimOE-ai runtime. The reader should have facility working with Linux and git at the command line.

What You'll Be Doing

This scenario builds on the work done in the previous scenario Scenario 2: Implementing a Two Agent Collection under mimOE-ai. As mentioned previously, this scenario is one in which there are three Agent Machines. Two of the Agent Machines were created in the previous scenario. The new Agent Machine is an AMD/ARM device that you will add to the mimik Service Mesh in this scenario. The figure below illustrates the physical architecture of this scenario.

Again, we assume that you've done the Two Agent Collection tutorial described previously and have the content and source code cloned from the mimOE-ai-developer-guide repo GitHub Repo.

Step 1: Go to the working directory created in the previous mimOE-ai tutorial. We assume this directory to be ./mimOE-ai/mimOE-ai-developer-guide. This directory contains the content and code for the various mimOE-ai tutorials. Type the following command in the terminal:

 cd ./mimOE-ai/mimOE-ai-developer-guide

Step 2: Follow the instructions in the mimOE-ai-developer-guide for a Three Agent Collection which is located in ReadMe file at:

 03-multi-agent-architecture/README.md

These instructions implement the Three Agent Collection scenario using a set of setup.http files that are executed in an instance of Visual Studio Code that has the REST Client Extension installed.