better doing

Bundled tasks hide AI gains.

Bundled tasks hide AI gains. “If you’re half as productive at debugging code you didn’t write, or less, the LLM saves you no time at all.” Philip Trammell’s point: automating one slice of a job can break the feedback loop […]

better doing

Better thinking & incentives

“The good news is, doing and thinking reinforce and augment each other. It’s a subtle but powerful feedback loop. You learn something. Armed with that new information, you do something. Informed by the results of your doing, you learn something […]