Real work, documented as it happens. Not demos, not theory—actual tasks where AI agents execute, iterate, and deliver. A running log of what works, what breaks, and what it means for how we work.
I’ve been using AI agents for real work—not experiments, not demos—for over a year now. The gap between what gets posted on X and what actually happens in practice is vast.
This series closes that gap.
Each entry documents an actual workflow: the task, the collaboration, the friction, the outcome. Some are mundane (video compression). Some are complex (multi-agent code refactors). All are real.
The Pattern:
I’m not proving AI works. I already know it does—for me, in my context, with my constraints. I’m documenting how it works so I can teach it, refine it, and see what patterns emerge over time.
The Arc (ongoing):
This isn’t a polished retrospective. It’s a running log. Some entries will be rough. That’s the point.
I needed to send 90 video files to a remote editor over mobile hotspot. The files were too large. What followed wasn't a tutorial lookup—it was a collaboration with an AI agent that analyzed, proposed, tested, pivoted, and executed. This is what agentic workflows actually look like.
Anthropic just shipped Claude Cowork—'Claude Code for the rest of your work.' It's the same agentic pattern I've been using for months, now packaged for non-developers. This isn't a product launch. It's a category shift in what AI assistance means.
I kept getting the same document wrong. Blog voice in architecture docs. Technical precision in executive briefs. Then I realized the problem wasn't AI—it was me. I was treating all documents as the same task.
I wrote a voice guide to help AI match my writing style. It worked too well—the AI learned the example phrases, not the principles behind them. Here's how I fixed it, and what it taught me about the difference between describing a voice and understanding one.
I had a 46-citation research paper about autonomous documentation. Academic frameworks rarely survive contact with a real codebase. So I asked an agent to turn theory into working code—and watched what happened.
The problem with AI coding assistants isn't capability—it's coordination. A single agent can write code. But who checks the work? Who remembers what was decided? I built a system where specialists implement and verifiers catch drift.
I built a Constitutional AI Governance Framework. Thirteen articles. HMAC attestations. Democratic amendment processes. Every validation function returned hardcoded perfection. The governance thinking was real. The code was theater. Here's what survived the extraction.