May 2026

Rotating Beers

The AI Hallucination Loop: Why Automation Still Needs a Human at the Wheel

Leave a Comment / Uncategorized /

We constantly hear that AI is going to automate everything, but what happens when the AI gets stuck in a loop of its own mistakes? When my partner’s restaurant lost their automated beer menu tool, I decided to build a replacement. What should have been a straightforward task turned into a frustrating battle with the “AI Hallucination Loop” across Claude and Gemini. Discover why automation still requires a human at the wheel, and learn three practical product management lessons to help you successfully build your own AI workflows without losing your mind.

The AI Hallucination Loop: Why Automation Still Needs a Human at the Wheel Read More »

Rebuilding the AI Agent: A Lesson in Strategy

Leave a Comment / Uncategorized /

Building with AI makes it easy to prioritize technical novelty over the actual problem. In this case study, I share the strategic lessons learned from rebuilding an AI content agent—shifting from a brittle, monolithic structure to four disciplined, modular workflows that prioritize educational quality and long-term flexibility over sheer content volume.

Rebuilding the AI Agent: A Lesson in Strategy Read More »

Building an AI Agent

Leave a Comment / Uncategorized /

So, I started over based on what I had learned firsthand, through my research, online reading, and talking to others. My first change was to break up my agent into several workflows instead of one long one. This makes troubleshooting a lot easier and gives me more flexibility as I learn more or tools change. I also want to change the focus, through targeted sources and prompts, so that my agent focuses on teaching, not the latest trend. Finally, I switched from 1min.ai to Google Gemini, as n8n works better with Gemini and causes fewer issues overall, letting me focus on refining the model rather than tracking down system errors.

Building an AI Agent Read More »