Flock of sheep grazing in a misty field

From Prompts to Agent Herds

Curiosity (2023)

When ChatGPT reached the world, I approached it the way an engineer approaches any new system: by taking it apart. In September 2023 I certified in prompt engineering, artificial intelligence and machine learning. Along the way I also explored what it took to run open models such as Llama 2 on my own servers.

My path with AI

A copilot in the editor (2024)

This was the year AI moved into my daily tools. It stopped being a website I visited and started to live inside my editor. I also began writing about where it was going, from deepfakes to autonomous security drones.

From suggestions to agents (2025)

Models stopped suggesting lines and started carrying out tasks. I learned Cursor, then n8n for self-hosted automation, and started experimenting with how to manage several AI development agents at once. The question changed from “what can it write?” to “what can I trust it to do?”

Building my own orchestration (2026)

After certifying in Claude Code, I stopped only using agents and started building systems around them. BetterAgentX coordinated specialized agents with automatic routing and persistent memory. PiCA explored approval workflows and code analysis. Every experiment taught the same lesson: agents are powerful, but they need boundaries, memory and human approval.

Today

My daily setup is Pi, a minimal coding agent, and Herdr, a terminal multiplexer that lets me watch a herd of agents at work. After 18+ years as an engineer, my job is still the same: understand the system, make it reliable, and stay responsible for the result.

Leave a Reply

Your email address will not be published. Required fields are marked *