Module 3: Platform AI to Custom Agents
Duration: 45 minutes Day: Day 1, Session 3
What This Module Is About
This module is the bridge. You've seen what platform AI can do (Module 2). You've felt what context engineering can do to a raw AI interaction (Module 1). Now you see what happens when you combine them: a custom agent that knows your infrastructure, follows your runbooks, and takes action through your tools.
The facilitator demos a live Hermes agent. Then you install it and run your first task.
Learning Objectives
By the end of this module, you will be able to:
- Observe a live domain agent analyzing the same CloudWatch alarm from Module 1 — but this time the agent reads the file, checks the runbook, and produces a structured diagnosis in one step
- Understand the gap between platform AI (stateless, service-scoped) and custom agents (stateful, context-rich, tool-enabled)
- Install the Hermes framework and run your first agent interaction
- Identify three operational tasks in your environment that would benefit from a domain agent
Prerequisites
- Module 1 complete (you understand context engineering)
- Module 2 complete (you understand the platform AI ceiling)
Module Contents
| Section | Content | Time |
|---|---|---|
| Demo | Hermes First-Run Walkthrough (facilitator-led) | 15 min |
| Lab | Install Hermes and Run Your First Task | 20 min |
| Reading | What Custom Agents Add | 10 min |
| Exploratory | Stretch Projects | Optional |
Key Insight: The Gap
Platform AI reacts to individual requests. Custom agents act on a persistent understanding of your environment.
The difference is the same as the difference between a generic contractor and an engineer who has been on your team for two years. The knowledge doesn't live in the questions you ask — it lives in the context they carry.
Hermes is the framework for encoding that context into an agent.