Field Notes
The ongoing investigation into AI-native company building.
Notes from research, implementation work, and the boundary between what AI can reliably own and what still needs a human. Following along is useful even if you aren't a client yet.
- AI OperationsHuman + AI Work
What should you never automate?
The interesting design question isn't what AI can do. It's where human involvement creates disproportionate value—and should stay that way.
2026-09-08
- AI OperationsHuman + AI Work
AI as infrastructure, not a chatbot
Why Workflowr exists: asking a question and copying the answer is a fraction of what's possible once AI is part of a company's operating architecture.
2026-09-01
Coming soon
Topics under active research. Not yet published—no invented content behind these titles.
- How far can one person scale with AI?Coming soon
- Skills vs. MCPs vs. agents: what actually belongs where?Coming soon
- I mapped every task required to launch a product. Here's what AI can own.Coming soon
- Building a marketing operation around ClaudeComing soon
- Why your AI stack probably has too much infrastructureComing soon
- The disappearing boundary between human and AI workComing soon
- Can one founder operate an entire marketing department?Coming soon
- What happens when Claude becomes part of the company instead of another app?Coming soon