🔗 ← Class 17: Data Pipelines and Workflow Automation | Class 19: Power Automate + Ask Sage API: AI-Augmented Workflow Automation →

image.png

<aside> 📥

Resources

📄 Lesson PDF

📊 Slides PDF

</aside>

Video Lesson

https://youtu.be/czkMmJzn6n8

Learning Objectives

image.png

By the end of this lesson, participants will be able to:

  1. Distinguish between a workflow (deterministic) and an agent (non-deterministic) and explain when each is appropriate
  2. Explain the Agent Studio 6 tool types and select the correct tool for a given task
  3. Design (on paper) an agent system prompt with persona, constraints, tool usage instructions, and HITL gates
  4. Explain how MCP enables the same agent architecture across Palantir, Ask Sage, and commercial tools
  5. Identify minimum security controls required before deploying a military AI agent

Introduction

In Class 16, you learned the 9 universal building blocks of automation. In Class 17, you applied those building blocks to data pipelines and workflow automation using Pipeline Builder and Power Automate. Both of those classes dealt with deterministic systems: you design the steps, the system executes them in order, every time. Now you cross a fundamental threshold. This class introduces systems that decide their own steps based on the input they receive. These systems are called agents.