10+ years ago, I used to spend much of my time writing ETL, debugging DAGs, and building dashboards. These days, I spend more of it directing agents, often doing work I wouldn’t have attempted a few years ago. I’m getting more done than ever.

I’ve come to think of agent modeling as one of the most important skills to develop: deciding what an agent is responsible for, what it knows, which tools it uses, and when it should ask for help. Then comes raising it, correcting its work, building trust, and sometimes asking it to keep *me* on track.

Drawing on my experience creating Apache Airflow and Apache Superset and working with agents at Preset, I’ll share what has worked, what hasn’t, and why doing this alone only gets you so far. We’ll talk about Multiplayer AI, shared ownership of agent workflows, and learning from how your colleagues work with AI. I’ll also introduce Agor, the BSL-licensed workbench we’re building to bring teams and agents together.