The AI Engineer Roadmap: Build Your First AI Agent

Several people have asked me for a roadmap to becoming an AI Engineer.

I do not think there is one roadmap. The field is moving too fast, the tools
are changing every week, and different people are coming from different places:
software, data, DevOps, automation, IT, product, and research.

But I do think there are foundational patterns.

One of them is this:

Build a small AI agent yourself.

Not a framework. Not a giant architecture diagram. Not a black box.

In this livestream, we will build a simple AI agent in less than 100 lines of
PowerShell and use it to explain the core pattern behind many agent systems:

- take a task from the user
- ask a model what to do next
- run a tool or command
- feed the result back to the model
- repeat until the task is done

We will start with the smallest useful version and then look at how the pattern
evolves:

- a model that suggests a PowerShell command
- an agent loop that can observe results
- structured tools for reading files, writing files, and running commands
- basic permission checks so the agent does not quietly do risky things

This is not about pretending 100 lines of PowerShell replaces production agent
frameworks. It is about understanding the moving parts well enough that those
frameworks stop feeling mysterious.

If you want to become an AI Engineer, this is one practical place to start:
build the loop, watch it work, and learn where the sharp edges are.

After the public livestream, I will continue the deep dive inside Agentic AI
Forge:

https://www.skool.com/agentic-ai-forge

In that working session, we will push the agent further, talk through the design
choices, and look at what it would take to turn the demo into a more serious
workflow.

About
Doug Finke is a 16-time Microsoft MVP, author of PowerShell for Developers, and creator of ImportExcel, PSAI, and PSClaudeCode. His work focuses on practical agentic AI: using scripts, shells, CLIs, and AI tools to build repeatable automation.